*653Mr. Justice Rehnquist

delivered the opinion of the Court.

Respondent John Jordan filed a complaint in the United States District Court for the Northern District of Illinois, individually and as a representative of a class, seeking declaratory and injunctive relief against two former directors of the Illinois Department of Public Aid, the director of the Cook County Department of Public Aid, and the comptroller of Cook County. Respondent alleged that these state officials were administering the federal-state programs of Aid to the Aged, Blind, or Disabled (AABD) in a manner inconsistent with various federal regulations and with the Fourteenth Amendment to the Constitution.1

AABD is one of the categorical aid programs administered by the Illinois Department of Public Aid pursuant to the Illinois Public Aid Code, Ill. Rev. Stat., c. 23, §§ 3-1 through 3-12 (1973). Under the Social Security Act, the program is funded by the State and the Federal Governments. 42 U. S. C. §§ 1381-1385.2 The Department of Health, Education, and Welfare (HEW), *654which administers these payments for the Federal Government, issued regulations prescribing maximum permissible time standards within which States participating in the program had to process AABD applications. Those regulations, originally issued in 1968, required, at the time of the institution of this suit, that eligibility determinations must be made by the States within 30 days of receipt of applications for aid to the aged and blind, and within 45. days of receipt of applications for aid to the disabled. For those persons found eligible, the assistance check was required to be received by them within the applicable time period. 45 CFR §206.10 (a)(3).3

*655During the period in which the federal regulations went into effect, Illinois public aid officials were administering the benefits pursuant to their own regulations as provided in the Categorical Assistance Manual of the Illinois Department of Public Aid.4 Respondent’s complaint charged that the Illinois defendants, operating under those regulations, were improperly authorizing grants to commence only with the month in which an application was approved and not including prior eligibility months for which an applicant was entitled to aid under federal law. The complaint also alleged that the Illinois defendants were not processing the applications within the applicable time requirements of the federal regulations; specifically, respondent alleged that his own application *656for disability benefits was not acted on by the Illlinois Department of Public Aid for almost four months. Such actions of the Illinois officials were alleged to violate federal law and deny the equal protection of the laws. Respondent's prayer requested declaratory and injunctive relief, and specifically requested “a permanent injunction enjoining the defendants to award to the entire class of plaintiffs all AABD benefits wrongfully withheld.”

In its judgment of March 15, 1972, the District Court declared § 4004 of the Illinois Manual to be invalid insofar as it was inconsistent with the federal regulations found in 45 CFR § 206.10 (a) (3), and granted a permanent injunction requiring compliance with the federal time limits for processing and paying AABD applicants. The District Court, in paragraph 5 of its judgment, also ordered the state officials to “release and remit AABD benefits wrongfully withheld to all applicants for AABD in the State of Illinois who applied between July 1, 1968 [the date of the federal regulations] and April 16, 197 [1] [the date of the preliminary injunction issued by the District Court] and were determined eligible . ...” 5

*657On appeal to the United States Court of Appeals for the Seventh Circuit, the Illinois officials contended, inter alia, that the Eleventh Amendment barred the award of *658retroactive benefits, that the judgment of inconsistency between the federal regulations and the provisions of the Illinois Categorical Assistance Manual could be given prospective effect only, and that the federal regulations in question were inconsistent with the Social Security Act itself. The Court of Appeals rejected these contentions and affirmed the judgment of the District Court. Jordan v. Weaver, 472 F. 2d 985 (1973).6 Because of an apparent conflict on the Eleventh Amendment issue with the decision of the Court of Appeals for the Second Circuit in Rothstein v. Wyman, 467 F. 2d 226 (1972), cert. denied, 411 U. S. 921 (1973), we granted the petition for certiorari filed by petitioner Joel Edelman, who is the present Director of the Illinois Department of Public Aid, and successor to the former directors sued below. 412 U. S. 937 (1973). The petition for certiorari raised the same contentions urged by the petitioner in the Court of Appeals.7 Because we believe the Court of Appeals *659erred in its disposition of the Eleventh Amendment claim, we reverse that portion of the Court of Appeals decision which affirmed the District Court’s order that retroactive benefits be paid by the Illinois state officials.8

*660The historical basis of the Eleventh Amendment has been oft stated, and it represents one of the more dramatic examples of this Court’s effort to derive meaning from the document given to the Nation by the Framers nearly 200 years ago. A leading historian of the Court tells us:

“The right of the Federal Judiciary to summon a State as defendant and to adjudicate its rights and liabilities had been the subject of deep apprehension and of active debate at the time of the adoption of the Constitution; but the existence of any such right had been disclaimed by many of the most eminent advocates of the new Federal Government, and it was largely owing to their successful dissipation of the fear of the existence of such Federal power that the Constitution was finally adopted.” 1 C. Warren, The Supreme Court in United States History 91 (rev. ed. 1937).

Despite such disclaimers,9 the very first suit entered *661in this Court at its February Term in 1791 was brought against the State of Maryland by a firm of Dutch bankers as creditors. Vanstophorst v. Maryland, see 2 Dall. *662401 and Warren, supra, at 91 n. 1. The subsequent year brought the institution of additional suits against other States, and caused considerable alarm and consternation in the country.

The issue was squarely presented to the Court in a suit brought at the August 1792 Term by two citizens of South Carolina, executors of a British creditor, against the State of Georgia. After a year's postponement for preparation on the part of the State of Georgia, the Court, after argument, rendered in February 1793, its short-lived decision in Chisholm v. Georgia, 2 Dall. 419. The decision in that case, that a State was liable to suit by a citizen of another State or of a foreign country, literally shocked the Nation. Sentiment for passage of a constitutional amendment to override the decision rapidly gained momentum, and five years after Chisholm the Eleventh Amendment was officially announced by President John Adams. Unchanged since then, the Amendment provides:

“The judicial power of the United States shall not be construed to extend to any suit in law or equity, commenced or prosecuted against one of the United States by Citizens of another State, or by Citizens or Subjects of any Foreign State.”

While the Amendment by its terms does not bar suits against a State by its own citizens, this Court has con*663sistently held that an unconsenting State is immune from suits brought in federal courts by her own citizens as well as by citizens of another State. Hans v. Louisiana, 134 U. S. 1 (1890); Duhne v. New Jersey, 251 U. S. 311 (1920); Great Northern Life Insurance Co. v. Read, 322 U. S. 47 (1944); Barden v. Terminal R. Co., 377 U. S. 184 (1964); Employees v. Department of Public Health and Welfare, 411 U. S. 279 (1973). It is also well established that even though a State is not named a party to the action, the suit may nonetheless be barred by the Eleventh Amendment. In Ford Motor Co. v. Department of Treasury, 323 U. S. 459 (1945), the Court said:

“[W]hen the action is in essence one for the recovery of money from the state, the state is the real, substantial party in interest and is entitled to invoke its sovereign immunity from suit even though individual officials are nominal defendants.” Id., at 464.

Thus the rule has evolved that a suit by private parties seeking to impose a liability which must be paid from public funds in the state treasury is barred by the Eleventh Amendment. Great Northern Life Insurance Co. v. Read, supra; Kennecott Copper Corp. v. State Tax Comm’n, 327 U. S. 573 (1946).

The Court of Appeals in this case, while recognizing that the Hans line of cases permitted the State to raise the Eleventh Amendment as a defense to suit by its own citizens, nevertheless concluded that the Amendment did not bar the award of retroactive payments of the statutory benefits found to have been wrongfully withheld. The Court of Appeals held that the above-cited cases, when read in light of this Court's landmark decision in Ex parte Young, 209 U. S. 123 (1908), do not preclude *664the grant of such a monetary award in the nature of equitable restitution.

Petitioner concedes that Ex parte Young, supra, is no bar to that part of the District Court’s judgment that prospectively enjoined petitioner’s predecessors from failing to process applications within the time limits established by the federal regulations. Petitioner argues, however, that Ex parte Young does not extend so far as to permit a suit which seeks the award of an accrued monetary liability which must be met from the general revenues of a State, absent consent or waiver by the State of its Eleventh Amendment immunity, and that therefore the award of retroactive benefits by the District Court was improper.

Ex parte Young was a watershed case in which this Court held that the Eleventh Amendment did not bar an action in the federal courts seeking to enjoin the Attorney General of Minnesota from enforcing a statute claimed to violate the Fourteenth Amendment of the United States Constitution. This holding has permitted the Civil War Amendments to the Constitution to serve as a sword, rather than merely as a shield, for those whom they were designed to protect. But the relief awarded in Ex parte Young was prospective only; the Attorney General of Minnesota was enjoined to conform his future conduct of that office to the requirement of the Fourteenth Amendment. Such relief is analogous to that awarded by the District Court in the prospective portion of its order under review in this case.

But the retroactive portion of the District Court’s order here, which requires the payment of a very substantial amount of money which that court held should have been paid, but was not, stands on quite a different footing. These funds will obviously not be paid out of the pocket of petitioner Edelman. Addressing himself to a similar situation in Rothstein v. Wyman, 467 F. 2d 226 *665(CA2 1972), cert. denied, 411 U. S. 921 (1973), Judge McGowan10 observed for the court:

“It is not pretended that these payments are to come from the personal resources of these appellants. Appellees expressly contemplate that they will, rather, involve substantial expenditures from the public funds of the state....

“It is one thing to tell the Commissioner of Social Services that he must comply with the federal standards for the future if the state is to have the benefit of federal funds in the programs he administers. It is quite another thing to order the Commissioner to use state funds to make reparation for the past. The latter would appear to us to fall afoul of the Eleventh Amendment if that basic constitutional provision is to be conceived of as having any present force.” 467 F. 2d, at 236-237 (footnotes omitted).

We agree with Judge McGowan’s observations. The funds to satisfy the award in this case must inevitably come from the general revenues of the State of Illinois, and thus the award resembles far more closely the monetary award against the State itself, Ford Motor Co. v. Department of Treasury, supra, than it does the prospective injunctive relief awarded in Ex parte Young.

The Court of Appeals, in upholding the award in this case, held that it was permissible because it was in the form of “equitable restitution” instead of damages, and therefore capable of being tailored in such a way as to minimize disruptions of the state program of categorical assistance. But we must judge the award actually made in this case, and not one which might have been differently tailored in a different case, and we must judge *666it in the context of the important constitutional principle embodied in the Eleventh Amendment.11

We do not read Ex parte Young or subsequent holdings of this Court to indicate that any form of relief may be awarded against a state officer, no matter how closely it may in practice resemble a money judgment payable out of the state treasury, so long as the relief may be labeled “equitable” in nature. The Court’s opinion in Ex parte Young hewed to no such line. Its citation of Hagood v. Southern, 117 U. S. 52 (1886), and In re Ayers, 123 U. S. 443 (1887), which were both actions *667against state officers for specific performance of a contract to which the State was a party, demonstrate that equitable relief may be barred by the Eleventh Amendment.

As in most areas of the law, the difference between the type of relief barred by the Eleventh Amendment and that permitted under Ex parte Young will not in many instances be that between day and night. The injunction issued in Ex parte Young was not totally without effect on the State's revenues, since the state law which the Attorney General was enjoined from enforcing provided substantial monetary penalties against railroads which did not conform to its provisions. Later cases from this Court have authorized equitable relief which has probably had greater impact on state treasuries than did that awarded in Ex parte Young. In Graham v. Richardson, 403 U. S. 365 (1971), Arizona and Pennsylvania welfare officials were prohibited from denying welfare benefits to otherwise qualified recipients who were aliens. In Goldberg v. Kelly, 397 U. S. 254 (1970), New York City welfare officials were enjoined from following New York State procedures which authorized the termination of benefits paid to welfare recipients without prior hearing.12 But the fiscal consequences to state *668treasuries in these cases were the necessary result of compliance with decrees which by their terms were prospective in nature. State officials, in order to shape their official conduct to the mandate of the Court’s decrees, would more likely have to spend money from the state treasury than if they had been left free to pursue their previous course of conduct. Such an ancillary effect on the state treasury is a permissible and often an inevitable consequence of the principle announced in Ex parte Young, supra.

But that portion of the District Court’s decree which petitioner challenges on Eleventh Amendment grounds goes much further than any of the cases cited. It requires payment of state funds, not as a necessary consequence of compliance in the future with a substantive federal-question determination, but as a form of compensation to those whose applications were processed on the slower time schedule at a time when petitioner was under no court-imposed obligation to conform to a different standard. While the Court of Appeals described this retroactive award of monetary relief as a form of “equitable restitution,” it is in practical effect indistinguishable in many aspects from an award of damages against the State. It will to a virtual certainty be paid from state funds, and not from the pockets of the individual state officials who were the defendants in the action. It is measured in terms of a monetary loss resulting from a past breach of a legal duty on the part of the defendant state officials.

Were we to uphold this portion of the District Court’s decree, we would be obligated to overrule the Court’s holding in Ford Motor Co. v. Department of Treasury, supra. There a taxpayer, who had, under protest, paid taxes to the State of Indiana, sought a refund of those taxes from the Indiana state officials who were charged with their collection. The taxpayer claimed that the tax *669had been imposed in violation of the United States Constitution. The term “equitable restitution” would seem even more applicable to the relief sought in that case, since the taxpayer had at one time had the money, and paid it over to the State pursuant to an allegedly unconstitutional tax exaction. Yet this Court had no hesitation in holding that the taxpayer’s action was a suit against the State, and barred by the Eleventh Amendment. We reach a similar conclusion with respect to the retroactive portion of the relief awarded by the District Court in this case.

The Court of Appeals expressed the view that its conclusion on the Eleventh Amendment issue was supported by this Court’s holding in Department of Employment v. United States, 385 U. S. 355 (1966). There the United States was held entitled to sue the Colorado Department of Employment in the United States District Court for refund of unemployment compensation taxes paid under protest by the American National Red Cross, an instrumentality of the United States. The discussion of the State’s Eleventh Amendment claim is confined to the following sentence in the opinion:

“With respect to appellants’ contention that the State of Colorado has not consented to suit in a federal forum even where the plaintiff is the United States, see Monaco v. Mississippi, 292 U. S. 313 (1934), and Ex parte Young, 209 U. S. 123 (1908).” Id., at 358.

Monaco v. Mississippi, 292 U. S. 313 (1934), reaffirmed the principle that the Eleventh Amendment was no bar to a suit by the United States against a State. Id., at 329. In view of Mr. Chief Justice Hughes’ vigorous reaffirmation in Monaco of the principles of the Eleventh Amendment and sovereign immunity, we think it unlikely that the Court in Department of Employment v. United States, in citing Ex parte Young as well as Monaco, *670intended to foreshadow a departure from the rule to which we adhere today.

Three fairly recent District Court judgments requiring state directors of public aid to make the type of retroactive payment involved here have been summarily affirmed by this Court notwithstanding Eleventh Amendment contentions made by state officers who were appealing from the District Court judgment.13 Shapiro v. Thompson, 394 U. S. 618 (1969), is the only instance in which the Eleventh Amendment objection to such retroactive relief was actually presented to this Court in a case which was orally argued. The three-judge District Court in that case had ordered the retroactive payment of welfare benefits found by that court to have been unlawfully withheld because of residence requirements held viola-tive of equal protection. 270 F. Supp. 331, 338 n. 5 (Conn. 1967). This Court, while affirming the judgment, did not in its opinion refer to or substantively treat the Eleventh Amendment argument. Nor, of course, did the summary dispositions of the three District Court cases contain any substantive discussion of this or any other issues raised by the parties.

This case, therefore, is the first opportunity the Court has taken to fully explore and treat the Eleventh Amend*671ment aspects of such relief in a written opinion. Shapiro v. Thompson and these three summary affirmances obviously are of precedential value in support of the contention that the Eleventh Amendment does not bar the relief awarded by the District Court in this case. Equally obviously, they are not of the same precedential value as would be an opinion of this Court treating the question on the merits. Since we deal with a constitutional question, we are less constrained by the principle of stare decisis than we are in other areas of the law.14 Having now had an opportunity to more fully consider the Eleventh Amendment issue after briefing and argument, we disapprove the Eleventh Amendment holdings of those cases to the extent that they are inconsistent with our holding today.

The Court of Appeals held in the alternative that even if the Eleventh Amendment be deemed a bar to the retroactive relief awarded respondent in this case, the State of Illinois had waived its Eleventh Amendment immunity and consented to the bringing of such a suit by participating in the federal AABD program. The Court of Appeals relied upon our holdings in Parden v. Terminal R. Co., 377 U. S. 184 (1964), and Petty v. Tennessee-Missouri Bridge Comm’n, 359 U. S. 275 (1959), *672and on the dissenting opinion of Judge Bright in Employees v. Department of Public Health and Welfare, 452 F. 2d 820, 827 (CA8 1971). While the holding in the latter case was ultimately affirmed by this Court in 411 U. S. 279 (1973), we do not think that the answer to the waiver question turns on the distinction between Parden, supra, and Employees, supra. Both Parden and Employees involved a congressional enactment which by its terms authorized suit by designated plaintiffs against a general class of defendants which literally included States or state instrumentalities. Similarly, Petty v. Tennessee-Missouri Bridge Comm’n, supra, involved congressional approval, pursuant to the Compact Clause, of a compact between Tennessee and Missouri, which provided that each compacting State would have the power “to contract, to sue, and be sued in its own name." The question of waiver or consent under the Eleventh Amendment was found in those cases to turn on whether Congress had intended to abrogate the immunity in question, and whether the State by its participation in the program authorized by Congress had in effect consented to the abrogation of that immunity.

But in this case the threshold fact of congressional authorization to sue a class of defendants which literally includes States is wholly absent. Thus respondent is not only precluded from relying on this Court’s holding in Employees, but on this Court’s holdings in Parden and Petty as well.15

*673The Court of Appeals held that as a matter of federal law Illinois had “constructively consented” to this suit by participating in the federal AABD program and agreeing to administer federal and state funds in compliance with federal law. Constructive consent is not a doctrine commonly associated with the surrender of constitutional rights, and we see no place for it here. In deciding whether a State has waived its constitutional protection under the Eleventh Amendment, we will find waiver only where stated “by the most express language or by such overwhelming implications from the text as [will] leave no room for any other reasonable construction.” Murray v. Wilson Distilling Co., 213 U. S. 151, 171 (1909). We see no reason to retreat from the Court's statement in Great Northern Life Insurance Co. v. Read, 322 U. S., at 54 (footnote omitted):

“[W]hen we are dealing with the sovereign exemption from judicial interference in the vital field of financial administration a clear declaration of the state’s intention to submit its fiscal problems to other courts than those of its own creation must be found.”

The mere fact that a State participates in a program through which the Federal Government provides assistance for the operation by the State of a system of public aid is not sufficient to establish consent on the part of the State to be sued in the federal courts. And while this Court has, in cases such as J. I. Case Co. v. Borak, 377 *674U. S. 426 (1964), authorized suits by one private party against another in order to effectuate a statutory purpose, it has never done so in the context of the Eleventh Amendment and a state defendant. Since Employees, supra, where Congress had expressly authorized suits against a general class of defendants and the only thing left to implication was whether the described class of defendants included States, was decided adversely to the putative plaintiffs on the waiver question, surely this respondent must also fail on that issue. The only language in the Social Security Act which purported to provide a federal sanction against a State which did not comply with federal requirements for the distribution of federal monies was found in former 42 U. S. C. § 1384 (now replaced by substantially similar provisions in 42 U. S. C. § 804), which provided for termination of future allocations of federal funds when a participating State failed to conform with federal law.16 This provision by its terms did not authorize suit against anyone, and standing alone, fell far short of a waiver by a participating State of its Eleventh Amendment immunity.

Our Brother Marshall argues in dissent, and the Court of Appeals held, that although the Social Security Act itself does not create a private cause of action, the cause of action created by 42 U. S. C. § 1983, coupled with the enactment of the AABD program, and the issuance by HEW of regulations which require the States to make corrective payments after successful “fair hear*675ings” and provide for federal matching funds to satisfy federal court orders of retroactive payments, indicate that Congress intended a cause of action for public aid recipients such as respondent.17 It is, of course, true that Rosado v. Wyman, 397 U. S. 397 (1970), held that suits in federal court under § 1983 are proper to secure compliance with the provisions of the Social Security Act on the part of participating States.18 But it has not hereto*676fore been suggested that § 1983 was intended to create a waiver of a State’s Eleventh Amendment immunity merely because an action could be brought under that *677section against state officers, rather than against the State itself. Though a § 1983 action may be instituted by public aid recipients such as respondent, a federal court’s remedial power, consistent with the Eleventh Amendment, is necessarily limited to prospective injunc-tive relief, Ex parte Young, supra, and may not include a retroactive award which requires the payment of funds from the state treasury, Ford Motor Co. v. Department of Treasury, supra.

Respondent urges that since the various Illinois officials sued in the District Court failed to raise the Eleventh Amendment as a defense to the relief sought by respondent, petitioner is therefore barred19 from raising the Eleventh Amendment defense in the Court of Appeals or in this Court. The Court of Appeals apparently felt the defense was properly presented, and dealt with it on the merits. We approve of this resolution, since it has been well settled since the decision *678in Ford Motor Co. v. Department of Treasury, supra, that the Eleventh Amendment defense sufficiently partakes of the nature of a jurisdictional bar so that it need not be raised in the trial court:

“[The Attorney General of Indiana] appeared in the federal District Court and the Circuit Court of Appeals and defended the suit on the merits. The objection to petitioner’s suit as a violation of the Eleventh Amendment was first made and argued by Indiana in this Court. This was in time, however. The Eleventh Amendment declares a policy and sets forth an explicit limitation on federal judicial power of such compelling force that this Court will consider the issue arising under this Amendment in this case even though urged for the first time in this Court.” 323 U. S., at 466-467.

For the foregoing reasons we decide that the Court of Appeals was wrong in holding that the Eleventh Amendment did not constitute a bar to that portion of the District Court decree which ordered retroactive payment of benefits found to have been wrongfully withheld. The judgment of the Court of Appeals is therefore reversed and the cause remanded for further proceedings consistent with this opinion.

So ordered.

Mr. Justice Douglas,

dissenting.

Congress provided in 42 U. S. C. § 1983 that:

“Every person who, under color of any statute, ordinance, regulation, custom, or usage, of any State or Territory, subjects, or causes to be subjected, any citizen of the United States or other person within the jurisdiction thereof to the deprivation of any rights, privileges, or immunities secured by the Con*679stitution and laws, shall be liable to the party injured in an action at law, suit in equity, or other proper proceeding for redress.”

In this class action respondent sought to enforce against state aid officials of Illinois provisions of the Social Security Act, 42 U. S. C. §§ 1381-1385, known as the Aid to the Aged, Blind, or Disabled (AABD) program.1 The complaint alleges violations of the Equal Protection Clause of the Fourteenth Amendment and also violations of the Social Security Act. Hence § 1983 is satisfied in haec verba, for a deprivation of “rights” which are “secured by the Constitution and laws” is alleged. The Court of Appeals, though ruling that the alleged constitutional violations had not occurred, sustained federal jurisdiction because federal “rights” were violated. The main issue tendered us is whether that ruling of the Court of Appeals is consistent with the Eleventh Amendment.2

*680Once the federal court had jurisdiction over the case, the fact that it ruled adversely to the claimant on the constitutional claim did not deprive it of its pendent jurisdiction over the statutory claim. United States v. Georgia Pub. Serv. Comm’n, 371 U. S. 285, 287-288.

In Ex parte Young, 209 U. S. 123, a suit by stockholders of a railroad was brought in a federal court against state officials to enjoin the imposition of confiscatory rates on the railroad in violation of the Fourteenth Amendment. The Eleventh Amendment was interposed as a defense. The Court rejected the defense, saying that state officials with authority to enforce state laws “who threaten and are about to commence proceedings, either of a civil or criminal nature, to enforce against parties affected an unconstitutional act, violating the Federal Constitution, may be enjoined by a Federal court of equity from such action.” Id., at 156. The Court went on to say that a state official seeking to enforce in the name of a State an unconstitutional act “comes into conflict with the superior authority of that Constitution, and he is in that case stripped of his official or representative character and is subjected in his person to the consequence of his individual conduct. The State has no power to impart to him any immunity from responsibility to the supreme authority of the United States.” Id., at 159-160.

As the complaint in the instant case alleges violations by officials of Illinois of the Equal Protection Clause of the Fourteenth Amendment, it seems that the case is governed by Ex parte Young so far as injunctive relief is concerned. The main thrust of the argument is that the. instant case asks for relief which if granted would .affect the treasury of the State.

Most welfare decisions by federal courts have a fi*681nancial impact on the States. Under the existing federal-state cooperative system, a state desiring to participate, submits a “state plan” to HEW for approval; once HEW approves the plan the State is locked into the cooperative scheme until it withdraws,3 all as described in King v. Smith, 392 U. S. 309, 316 et seq. The welfare cases coming here have involved ultimately the financial responsibility of the State to beneficiaries claiming they were deprived of federal rights. King v. Smith required payment to children even though their mother was co-habitating with a man who could not pass muster as a *682“parent.” Rosado v. Wyman, 397 U. S. 397, held that under this state-federal cooperative program a State could not reduce its standard of need in conflict with the federal standard. It is true that Rosado did not involve retroactive payments as are involved here. But the distinction is not relevant or material because the result in every welfare case coming here is to increase or reduce the financial responsibility of the participating State. In no case when the responsibility of the State is increased to meet the lawful demand of the beneficiary, is there any levy on state funds. Whether the decree is prospective only or requires payments for the weeks or months wrongfully skipped over by the state officials, the nature of the impact on the state treasury is precisely the same.

We have granted relief in other welfare cases which included retroactive assistance benefits or payments. In State Dept. of Health and Rehabilitative Services v. Zarate, 407 U. S. 918, the sole issue presented to us4 was whether the Eleventh Amendment barred a judgment against state officers for retroactive welfare assistance benefits or payments. That had been ordered by the lower court and we summarily affirmed, only Mr. Justice White voting to note probable jurisdiction. We also summarily affirmed the judgment in Sterrett v. Mothers’ & Children’s Rights Organization, 409 U. S. 809, where one of the two questions 5 was whether retroactive payments of benefits violated the Eleventh Amendment. In Wyman v. Bowens, 397 U. S. 49, we affirmed a judgment *683where payments were awarded in spite of the argument that the order was an incursion on the Eleventh Amendment.6 In Shapiro v. Thompson, 394 U. S. 618, we affirmed a judgment which ordered payment of benefits wrongfully withheld;7 and while we did not specifically refer to the point, the lower court had expressly rejected the Eleventh Amendment argument.8

In Gaither v. Sterrett, 346 F. Supp. 1095, 1099, whose judgment we affirmed,9 409 U. S. 1070, the court said:

“[T]his court would note that if defendants’ position regarding the jurisdictional bar of the Eleventh Amendment is correct, a great number of federal district court judgments are void, and the Supreme Court has affirmed many of these void judgments.”

The Court of Appeals for the Seventh Circuit is in line with that view; the opposed view of the Court of Appeals for the Second Circuit in Rothstein v. Wyman, 467 F. 2d 226, is out of harmony with the established law.

What is asked by the instant case is minor compared to the relief granted in Griffin v. School Board, 377 U. S. 218. In that case we authorized entry of an order putting an end to a segregated school system. We held, inter alia, that “the District Court may, if necessary to prevent further racial discrimination, require the Supervisors to *684exercise the power that is theirs to levy taxes to raise funds adequate to reopen, operate, and maintain without racial discrimination a public school system in Prince Edward County like that operated in other counties in Virginia.” Id., at 233. We so held against vigorous contentions of the state officials that the Eleventh Amendment protected the State; and in reply we cited Lincoln County v. Luning, 133 U. S. 529, and Kennecott Copper Corp. v. State Tax Comm’n, 327 U. S. 573, 579, to support the proposition that “actions against a county can be maintained in United States courts in order to vindicate federally guaranteed rights.” 377 U. S., at 233.

Griffin is sought to be distinguished on the ground that a “county” is not the “state” for purposes of the Eleventh Amendment. But constitutionally the county in Griffin was exercising state policy as are the counties here, because otherwise the claim of denial of equal protection would be of no avail.

Counties are citizens of their State for purposes of diversity of citizenship. Bullard v. City of Cisco, 290 U. S. 179; Moor v. County of Alameda, 411 U. S. 693, 718-719. And they are not States for purposes of 28 U. S. C. § 1251 (a) which gives this Court original and exclusive jurisdiction of: “(1) All controversies between two or more states. . . .” Illinois v. City of Milwaukee, 406 U. S. 91, 98. But, being citizens of their State, suits against them by another State are in our original but not exclusive jurisdiction under 28 U. S. C. § 1251 (b)(3). Ibid. Yet, as agencies of the State whether in carrying out educational policies or otherwise, they are the State, as Griffin held, for purposes of the Fourteenth Amendment. And Griffin, like the present case, dealt only with liability to citizens for state policy and state action.

Yet petitioner asserts that money damages may not be awarded against state offenses, as such a judgment *685will expend itself on the state treasury. But we are unable to say that Illinois on entering the federal-state welfare program waived its immunity to suit for injunctions but did not waive its immunity for compensatory awards which remedy its willful defaults of obligations undertaken when it joined the cooperative venture.10

It is said however, that the Eleventh Amendment is concerned, not with immunity of States from suit, but with the jurisdiction of the federal courts to entertain the suit. The Eleventh Amendment does not speak of "jurisdiction”; it withholds the “judicial power” of federal courts “to any suit in law or equity . . . against one of the United States . . . .” If that “judicial power,” or “jurisdiction” if one prefers that concept, may not be exercised even in “any suit in . . . equity” then Ex parte Young should be overruled. But there is none eager to take the step. Where a State has consented to join a federal-state cooperative project, it is realistic to conclude that the State has agreed to assume its obligations under that legislation. There is nothing in the Eleventh Amendment to suggest a difference between suits at law and suits in equity, for it treats the two without distinction. If common sense has any role to play in constitutional adjudication, once there is a waiver of immunity it must be true that it is complete so far as effective operation of the state-federal joint welfare program is concerned.

*686We have not always been unanimous in concluding when a State has waived its immunity. In Parden v. Terminal R. Co., 377 U. S. 184, where Alabama was sued by some of its citizens for injuries suffered in the interstate operation of an Alabama railroad, the State defended on the grounds of the Eleventh Amendment. The Court held that Alabama was liable as a carrier under the Federal Employers' Liability Act, saying:

“Our conclusion is simply that Alabama, when it began operation of an interstate railroad approximately 20 years after enactment of the FELA, necessarily consented to such suit as was authorized by that Act,” id., at 192.

The Court added:

“Our conclusion that this suit may be maintained is in accord with the common sense of this Nation’s federalism. A State’s immunity from suit by an individual without its consent has been fully recognized by the Eleventh Amendment and by subsequent decisions of this Court. But when a State leaves the sphere that is exclusively its own and enters into activities subject to congressional regulation, it subjects itself to that regulation as fully as if it were a private person or corporation.” Id., at 196.

As the Court of Appeals in the instant case concluded, Illinois by entering into the joint federal-state welfare plan just as surely “ [left] the sphere that is exclusively its own.” Ibid.

It is argued that participation in the program of federal financial assistance is not sufficient to establish consent on the part of the State to be sued in federal courts. But it is not merely participation which supports a finding of Eleventh Amendment waiver, but *687participation in light of the existing state of the law as exhibited in such decisions as Shapiro v. Thompson, 394 U. S. 618, which affirmed judgments ordering retroactive payment of benefits. Today's holding that the Eleventh Amendment forbids court-ordered retroactive payments, as the Court recognizes, necessitates an express overruling of several of our recent decisions. But it was against the background of those decisions that Illinois continued its participation in the federal program, and it can hardly be claimed that such participation was in ignorance of the possibility of court-ordered retroactive payments. The decision to .participate against that background of precedent can only be viewed as a waiver of immunity from such judgments.

I would affirm the judgment of the Court of Appeals.

Mr. Justice Brennan,

dissenting.

This suit is brought by Illinois citizens against Illinois officials. In that circumstance, Illinois may not invoke the Eleventh Amendment, since that Amendment bars only federal court suits against States by citizens of other States. Rather, the question is whether Illinois may avail itself of the nonconstitutional but ancient doctrine of sovereign immunity as a bar to respondent's claim for retroactive AABD payments. In my view Illinois may not assert sovereign immunity for the reason I expressed in dissent in Employees v. Department of Public Health and Welfare, 411 U. S. 279, 298 (1973): the States surrendered that immunity in' Hamilton's words, “in the plan of the Convention,” that formed the Union, at least insofar as the States granted Congress specifically enumerated powers. See id., at 319 n. 7; Parden v. Terminal R. Co., 377 U. S. 184 (1964). Congressional authority to enact the Social Security Act, of which AABD is a part, former 42 U. S. C. §§ 1381-1385 *688(now replaced by similar provisions in 42 U. S. C. § 801-804 (1970 ed., Supp. II)), is to be found in Art. I, § 8, cl. 1, one of the enumerated powers granted Congress by the States in the Constitution. I remain of the opinion that “because of its surrender, no immunity exists that can be the subject of a congressional declaration or a voluntary waiver,” 411 U. S., at 300, and thus have no occasion to inquire whether or not Congress authorized an action for AABD retroactive benefits, or whether or not Illinois voluntarily waived the immunity by its continued participation in the program against the background of precedents which sustained judgments ordering retroactive payments.

I would affirm the judgment of the Court of Appeals.

Me. Justice Marshall,

with whom Mr. Justice Blackmun joins, dissenting.

The Social Security Act’s categorical assistance programs, including the Aid to the Aged, Blind, or Disabled (AABD) program involved here, are fundamentally different from most federal legislation. Unlike the Fair Labor Standards Act involved in last Term’s decision in Employees v. Department of Public Health and Welfare, 411 U. S. 279 (1973), or the Federal Employers’ Liability Act-at issue in Parden v. Terminal R. Co., 377 U. S. 184 (1964), the Social Security Act does not impose federal standards and liability upon all who engage in certain regulated activities, including often-unwilling state agencies. Instead, the Act seeks to induce state participation in the federal welfare programs by offering federal matching funds in exchange for the State’s voluntary assumption of the Act’s requirements. I find this basic distinction crucial: it leads me to conclude that by participation in the programs, the States waive whatever immunity they might otherwise have from federal court *689orders requiring retroactive payment of welfare benefits.1

In its contacts with the Social Security Act’s assistance programs in recent years, the Court has frequently described the Act as a “scheme of cooperative federalism.” See, e. g., King v. Smith, 392 U. S. 309, 316 (1968); Jefferson v. Hackney, 406 U. S. 535, 542 (1972). While this phrase captures a number of the unique characteristics of these programs, for present purposes it serves to emphasize that the States’ decision to participate in the programs is a voluntary one. In deciding to participate, however, the States necessarily give up their freedom to operate assistance programs for the needy as they see fit, and bind themselves to conform their programs to the requirements of the federal statute and regulations. As the Court explained in King v. Smith, supra, at 316-317 (citations omitted):

“States are not required to participate in the program, but those which desire to take advantage of the substantial federal funds available for distribution to needy children [or needy aged, blind or disabled] are required to submit an AFDC [or AABD] plan for the approval of the Secretary of Health, Education, and Welfare (HEW). The plan must conform with several requirements of the Social Security Act and with rules and regulations promulgated by HEW.”

So here, Illinois elected to participate in the AABD program, and received and expended substantial federal funds in the years at issue. It thereby obligated itself to comply with federal law, including the require*690ment of former 42 U. S. C. § 1382 (a) (8) that “such aid or assistance shall be furnished with reasonable promptness to all eligible individuals.” In Townsend v. Swank, 404 U. S. 282, 286 (1971), we held that participating States must strictly comply with the requirement that aid be furnished “to all eligible individuals,” and that the States have no power to impose additional eligibility requirements which exclude persons eligible for assistance under federal standards. Today’s decision, ante, at 659-660, n. 8, properly emphasizes that participating States must also comply strictly with the “reasonable promptness” requirement and the more detailed regulations'adding content to it.

In agreeing to comply with the requirements of the Social Security Act and HEW regulations, I believe that Illinois has also agreed to subject itself to suit in the federal courts to enforce these obligations. I recognize, of course, that the Social Security Act does not itself provide for a cause of action to enforce its obligations. As the Court points out, the only sanction expressly provided in the Act for a participating State’s failure to comply with federal requirements is the cutoff of federal funding by the Secretary of HEW. Former 42 U. S. C. § 1384 (now 42 U. S. C. § 804 (1970 ed., Supp. II)).

But a cause of action is clearly provided by 42 U. S. C. § 1983, which in terms authorizes suits to redress deprivations of rights secured by the “laws” of the United States. And we have already rejected the argument that Congress intended the funding cutoff to be the sole remedy for noncompliance with federal requirements. In Rosado v. Wyman, 397 U. S. 397, 420-423 (1970), we held that suits in federal court under § 1983 were proper to enforce the provisions of the Social Security Act against participating States. Mr. Justice Harlan, writing for the Court, ex*691amined the legislative history and found “not the slightest indication” that Congress intended to prohibit suits in federal court to enforce compliance with federal standards. Id,., at 422.

I believe that Congress also intended the full panoply of traditional judicial remedies to be available to the federal courts in these § 1983 suits. There is surely no indication of any congressional intent to restrict the courts’ equitable jurisdiction. Yet the Court has held that “[ujnless a statute in so many words, or by a necessary and inescapable inference, restricts the court’s jurisdiction in equity, the full scope of that jurisdiction is to be recognized and applied.” Porter v. Warner Holding Co., 328 U. S. 395, 398 (1946). “When Congress entrusts to an equity court the enforcement of prohibitions contained in a regulatory enactment, it must be taken to have acted cognizant of the historic power of equity to provide complete relief in light of the statutory purposes.” Mitchell v. DeMario Jewelry, 361 U. S. 288, 291-292 (1960).

In particular, I am firmly convinced that Congress intended the restitution of wrongfully withheld assistance payments to be a remedy available to.the federal courts in these suits. Benefits under the categorical assistance programs “are a matter of statutory entitlement for persons qualified to receive them.” Goldberg v. Kelly, 397 U. S. 254, 262 (1970). Retroactive payment of benefits secures for recipients this entitlement which was withheld in violation of federal law. Equally important, the courts’ power to order retroactive payments is an essential remedy to insure future state compliance with federal requirements. See Porter v. Warner Holding Co., supra, at 400. No other remedy can effectively deter States from the strong temptation to cut *692welfare budgets by circumventing the stringent requirements of federal law. The funding cutoff is a drastic sanction, one which HEW has proved unwilling or unable to employ to compel strict compliance with the Act and regulations. See Rosado v. Wyman, supra, at 426 (Douglas, J., concurring). Moreover, the cutoff operates only prospectively; it in no way deters the States from even a flagrant violation of the Act's requirements for as long as HEW does not discover the violation and threaten to take such action.

Absent any remedy which may act with retroactive effect, state welfare officials have everything to gain and nothing to lose by failing to comply with the congressional mandate that assistance be paid with reasonable promptness to all eligible individuals. This is not idle speculation without basis in practical experience. In this very case, for example, Illinois officials have knowingly violated since 1968 federal regulations on the strength of an argument as to its invalidity which even the majority deems unworthy of discussion. Ante, at 659-660, n. 8. Without a retroactive-payment remedy, we are indeed faced with “the spectre of a state, perhaps calculatingly, defying federal law and thereby depriving welfare recipients of the financial assistance Congress thought it was giving them.” Jordan v. Weaver, 472 F. 2d 985, 995 (CA7 1972). Like the Court of Appeals, I cannot believe that Congress could possibly have intended any such result.

Such indicia of congressional intent as can be gleaned from the statute confirm that Congress intended to authorize retroactive payment of assistance benefits unlawfully withheld. Availability of such payments is implicit in the “fair hearing” requirement, former 42 U. S. C. § 1382 (a) (4), which permitted welfare recipients to challenge the denial of assistance. The regulations *693which require States to make corrective payments retroactively in the event of a successful fair hearing challenge, 45 CFR § 205.10 (a) (18), merely confirm the obvious statutory intent. .HEW regulations also authorize federal matching funds for retroactive assistance payments made pursuant to court order, 45 CFR §§ 205.10 (b) (2), (b)(3). We should not lightly disregard this explicit recognition by the agency charged with administration of the statute that such a remedy was authorized by Congress. See Griggs v. Duke Power Co., 401 U. S. 424, 433-434 (1971).

Illinois chose to participate in the AABD program with its eyes wide open. Drawn by the lure of federal funds, it voluntarily obligated itself to comply with the Social Security Act and HEW regulations, with full knowledge that Congress had authorized assistance recipients to go into federal court to enforce these obligations and to recover benefits wrongfully denied. Any doubts on this score must surely have been removed by our decisions in Rosado v. Wyman, supra, and Shapiro v. Thompson, 394 U. S. 618 (1969), where we affirmed a district court retroactive payment order. I cannot avoid the conclusion that, by virtue of its knowing and voluntary decision to nevertheless participate in the program, the State necessarily consented to subject itself to these suits. I have no quarrel with the Court’s view that waiver of constitutional rights should not lightly be inferred. But I simply cannot believe that the State could have entered into this essentially contractual agreement with the Federal Government without recognizing that it was subjecting itself to the full scope of the § 1983 remedy provided by Congress to enforce the terms of the agreement.

Of course, § 1983 suits are nominally brought against state officers, rather than the State itself, and do not *694ordinarily raise Eleventh Amendment problems in view of this Court’s decision in Ex parte Young, 209 U. S. 123 (1908). But to the extent that the relief authorized by Congress in an action under § 1983 may be open to Eleventh Amendment objections,2 these objections are waived when the State agrees to comply with federal requirements enforceable in such an action. I do not find persuasive the Court’s reliance in this case on the fact that “congressional authorization to sue a class of defendants which literally includes States” is absent. Ante, at 672. While true, this fact is irrelevant here, for this is simply not a case “literally” against the State. While the Court successfully knocks down the strawman it has thus set up, it never comes to grips with the undeniable fact that Congress has “literally” authorized this suit within the terms of § 1983. Since there is every reason to believe that Congress intended the full panoply of judicial remedies to be available in § 1983 equitable actions to enforce the Social Security Act, I think the conclusion is inescapable that Congress authorized and the State consented to § 1983 actions in which the relief might otherwise be questioned on Eleventh Amendment grounds.

My conclusion that the State has waived its Eleventh Amendment objections to court-ordered retroactive assistance payments is fully consistent with last Term’s *695decision in Employees v. Department of Public Health and Welfare, 411 U. S. 279 (1973). As I emphasized in my concurring opinion, there was no voluntary action by the State in Employees which could reasonably be construed as evidencing its consent to suit in a federal forum.

“[T]he State was fully engaged in the operation of the affected hospitals and schools at the time of the 1966 amendments. To suggest that the State had the choice of either ceasing operation of these vital public services or 'consenting’ to federal suit suffices, I believe, to demonstrate that the State had no true choice at all and thereby that the State did not voluntarily consent to the exercise of federal jurisdiction . . . .” Id., at 296.

A finding of waiver here is also consistent with the reasoning of the majority in Employees, which relied on a distinction between “governmental” and “proprietary” functions of state government. Id., at 284-285. This distinction apparently recognizes that if sovereign immunity is to be at all meaningful, the Court must be reluctant to hold a State to have waived its immunity simply by acting in its sovereign capacity — i. e., by merely performing its “governmental” functions. On the other hand, in launching a profitmaking enterprise, “a State leaves the sphere that is exclusively its own,” Parden v. Terminal R. Co., 377 U. S., at 196, and a voluntary waiver of sovereign immunity can more easily be found. While conducting an assistance program for the needy is surely a “governmental” function, the State here has done far more than operate its own program in its sovereign capacity. It has voluntarily subordinated its sovereignty in this matter to that of the Federal Government, and agreed to comply with the conditions imposed *696by Congress upon the expenditure of federal funds. In entering this federal-state cooperative program, the State again “leaves the sphere that is exclusively its own,” and similarly may more readily be found to have voluntarily waived its immunity.

Indeed, this is the lesson to be drawn from this Court’s decision in Petty v. Tennessee-Missouri Bridge Comm’n, 359 U. S. 275 (1959), where the Court found that the States had waived the sovereign immunity of the Commission by joining in an interstate compact subject to the approval of Congress. The Court in Petty emphasized that it was “called on to interpret not unilateral state action but the terms of a consensual agreement” between the States and Congress, id., at 279; and held that the States who join such a consensual agreement, “by accepting it and acting under it assume the conditions that Congress under the Constitution attached.” Id., at 281-282. Although the congressional intent regarding the sue-and-be-sued clause was by no means certain, the Court held that the surrounding conditions made it clear that the States accepting it waived their sovereign immunity, id., at 280, especially since this interpretation was necessary to keep the compact “a living interstate agreement which performs high functions in our federalism.” Id., at 279.

I find the approach in Petty controlling here. As even the dissent in that case recognized, id., at 285 (Frankfurter, J., dissenting), Congress undoubtedly has the power to insist upon a waiver of sovereign immunity as a condition of its consent to such a federal-state agreement. Since I am satisfied that Congress has in fact done so here, at least to the extent that the federal courts may do “complete rather than truncated justice,” Porter v. Warner Holding Co., 328 U. S., at 398, in § 1983 actions authorized by Congress against state welfare authorities, I respectfully dissent.

*447Me. Justice Rehnquist

delivered the opinion of the Court.

In the 1972 Amendments to Title VII of the Civil Rights Act of 1964, Congress, acting under § 5 of the Fourteenth Amendment, authorized federal courts to award money damages in favor of a private individual against a state government found to have subjected that person to employment discrimination on the basis of *448“race, color, religion, sex, or national origin/'1 The principal question presented by these cases is whether, as against the shield of sovereign immunity afforded the State by the Eleventh Amendment, Edelman v. Jordan, 415 U. S. 651 (1974), Congress has the power to authorize federal courts to enter such an award against the State as a means of enforcing the substantive guarantees of the Fourteenth Amendment. The Court of Appeals for the Second Circuit held that the effect of our decision in Edelman was to foreclose Congress’ power. We granted certiorari to resolve this important constitutional question. 423 IT. S. 1031 (1975). We reverse.

I

Petitioners in No. 75-251 sued in the United States District Court for the District of Connecticut on behalf of all present and retired male employees of the State of Connecticut. Their amended complaint asserted, inter alia, that certain provisions in the State’s statutory retirement benefit plan discriminated against them because of their sex, and therefore contravened Title VII of the 1964 Act, 78 Stat. 253, as amended, 42 U. S. C. § 2000e et seq. (1970 ed. and Supp. IV). Title VII, which originally did not include state and local govern-*449merits, had in the interim been amended to bring the States within its purview.2

The District Court held that the Connecticut State Employees Retirement Act violated Title VII's prohibition against sex-based employment discrimination. 390 F. Supp. 278, 285-288 (1974).3 It entered prospective injunctive relief in petitioners’ favor against respondent state officials.4 Petitioners also sought an award of retroactive retirement benefits as compensation for losses *450caused by the State's discrimination,5 as well as “a reasonable attorney's fee as part of the costs.” 6 But the District Court held that both would constitute recovery of money damages from the State’s treasury, and were therefore precluded by the Eleventh Amendment and by this Court’s decision in Edelman v. Jordan, supra.

On petitioners’ appeal,7 the Court of Appeals affirmed in part and reversed in part. It agreed with the District Court that the action, “insofar as it seeks damages, is in essence against the state and as such is subject to the Eleventh Amendment.” 519 P. 2d 559, 565 (1975). The Court of Appeals also found that under the 1972 Amendments to Title VII, “Congress intended to authorize a private suit for backpay by state employees against the state.” Id., at 568. Notwithstanding this statutory authority, the Court of Appeals affirmed the District Court and held that under Edelman a “private federal action for retroactive damages” is not a “constitu*451tionally permissible method of enforcing Fourteenth Amendment rights.” 519 F. 2d, at 569. It reversed the District Court and remanded as to attorneys’ fees, however, reasoning that such an award would have only an “ancillary effect” on the state treasury of the kind permitted under Edelman, supra, at 667-668. 519 F. 2d, at 571. The petition filed here by the state employees in No. 75-251 contends that Congress does possess the constitutional power under § 5 of the Fourteenth Amendment to authorize their Title VII damages action against the State. The state officials’ cross-petition, No. 75-283, argues that under Edelman the Eleventh Amendment bars any award of attorneys’ fees here because it would be paid out of the state treasury.

II

In Edelman this Court held that monetary relief awarded by the District Court to welfare plaintiffs, by reason of wrongful denial of benefits which had occurred previous to the entry of the District Court’s determination of their wrongfulness, violated the Eleventh Amendment. Such an award was found to be indistinguishable from a monetary award against the State itself which had been prohibited in Ford Motor Co. v. Department of Treasury, 323 U. S. 459, 464 (1945). It was therefore controlled by that ease rather than by Ex parte Young, 209 U. S. 123 (1908), which permitted suits against state officials to obtain prospective relief against violations of the Fourteenth Amendment.

Edelman went on to hold that the plaintiffs in that case could not avail themselves of the doctrine of waiver expounded in cases such as Parden v. Terminal R. Co., 377 U. S. 184 (1964), and Employees v. Missouri Public Health Dept., 411 U. S. 279 (1973), because the necessary predicate for that doctrine was congres*452sional intent to abrogate the immunity conferred by the Eleventh Amendment. We concluded that none of the statutes relied upon by plaintiffs in Edelman contained any authorization by Congress to join a State as defendant. The Civil Rights Act of 1871, 42 U. S. C. § 1983, had been held in Monroe v. Pape, 365 U. S. 167, 187-191 (1961), to exclude cities and other municipal corporations from its ambit; that being the case, it could not have been intended to include States as parties defendant. The provisions of the Social Security Act relied upon by plaintiffs were held by their terms not to “authorize suit against anyone,” 415 U. S., at 674, and they, too, were incapable of supplying the predicate for a claim of waiver on the part of the State.

All parties in the instant litigation agree with the Court of Appeals that the suit for retroactive benefits by the petitioners is in fact indistinguishable from that sought to be maintained in Edelman, since what is sought here is a damages award payable to a private party from the state treasury.8

Our analysis begins where Edelman ended, for in this Title VII case the “threshold fact of congressional authorization,” id., at 672, to sue the State as employer is clearly present. This is, of course, the prerequisite found present in Farden and wanting in Employees. We are aware of the factual differences between the type of state activity involved in Farden and that involved in the present case, but we do not think that difference is material for our purposes. The congressional authorization involved in Parden was based on the power of Congress under the Commerce Clause; here, however, the *453Eleventh Amendment defense is asserted in the context of legislation passed pursuant to Congress’ authority under § 5 of the Fourteenth Amendment.9

As ratified by the States after the Civil War, that Amendment quite clearly contemplates limitations on their authority. In relevant part, it provides:

“Section 1. . . . No State shall make or enforce any law which shall abridge the privileges or immunities of citizens of the United States; nor shall any State deprive any person of life, liberty, or property, without due process of law; nor deny to any person within its jurisdiction the equal protection of the laws.

“Section 5. The Congress shall have power to enforce, by appropriate legislation, the provisions of this article.”

The substantive provisions are by express terms directed at the States. Impressed upon them by those provisions are duties with respect to their treatment of private individuals. Standing behind the imperatives is Congress’ power to “enforce” them “by appropriate legislation.”

The impact of the Fourteenth Amendment upon the relationship between the Federal Government and the States, and the reach of congressional power under § 5, were examined at length by this Court in Ex 'parte Virginia, 100 U. S. 339 (1880). A state judge had been arrested and indicted under a federal criminal statute *454prohibiting the exclusion on the basis of race of any citizen from, service as a juror in a state court. The judge claimed that the statute was beyond Congress’ power to enact under either the Thirteenth10 or the Fourteenth Amendment. The Court first observed that these Amendments “were intended to be, what they really are, limitations of the power of the States and enlargements of the power of Congress.” Id., at 345. It then addressed the relationship between the language of § 5 and the substantive provisions of the Fourteenth Amendment:

“The prohibitions of the Fourteenth Amendment are directed to the States, and they are to a degree restrictions of State power. It is these which Congress is empowered to enforce, and to enforce against State action, however put forth, whether that action be executive, legislative, or judicial. Such enforcement is no invasion of State sovereignty. No law can be, which the people of the States have, by the Constitution of the United States, empowered Congress to enact. ... It is said the selection of jurors for her courts and the administration of her laws belong to each State; that they are her rights. This is true in the general. But in exercising her rights, a State cannot disregard the limitations which the Federal Constitution has applied to her power. Her rights do not reach to that extent. Nor can she deny to the general government the right to exercise all its *455granted powers, though they may interfere with the full enjoyment of rights she would have if those powers had not been thus granted. Indeed, every addition of power to the general government involves a corresponding diminution of the governmental powers of the States. It is carved out of them.

“The argument in support of the petition for a habeas corpus ignores entirely the power conferred upon Congress by the Fourteenth Amendment. Were it not for the fifth section of that amendment, there might be room for argument that the first section is only declaratory of the moral duty of the State .... But the Constitution now expressly gives authority for congressional interference and compulsion in the cases embraced within the Fourteenth Amendment. It is but a limited authority, true, extending only to a single class of cases; but within its limits it is complete." Id., at 346-348.

Ex parte Virginia’s early recognition of this shift in the federal-state balance has been carried forward by more recent decisions of this Court. See, e: g., South Carolina v. Katsenbach, 383 U. S. 301, 308 (1966); Mitchum v. Foster, 407 U. S. 225, 238-239 (1972).

There can be no doubt that this line of cases has sanctioned intrusions by Congress, acting under the Civil War Amendments, into the judicial, executive, and legislative spheres of autonomy previously reserved to the States. The legislation considered in each case was grounded on the expansion of Congress’ powers— with the corresponding diminution of state sovereignty— found to be intended by the Framers and made part of the Constitution upon the States’ ratification of those *456Amendments, a phenomenon aptly described as a “carv[ing] out” in Ex parte Virginia, supra, at 346.

It is true that none of these previous cases presented the question of the relationship between- the Eleventh Amendment and the enforcement power granted to Congress under § 5 of the Fourteenth Amendment. But we think that the Eleventh Amendment, and the principle of state sovereignty which it embodies, see Hans v. Louisiana, 134 U. S. 1 (1890), are necessarily limited by the enforcement provisions of § 5 of the Fourteenth Amendment. In that section Congress is expressly granted authority to enforce “by appropriate legislation” the substantive provisions of the Fourteenth Amendment, which themselves embody significant limitations on state authority. When Congress acts pursuant to § 5, not only is it exercising legislative authority that is plenary within the terms of the constitutional grant, it is exercising that authority under one section of a constitutional Amendment whose other sections by their own terms embody limitations on state authority. We think that Congress may, in determining what is “appropriate legislation” for the purpose of enforcing the provisions of the Fourteenth Amendment, provide for private suits against States or state officials which are constitutionally impermissible in other contexts.11 See Edelman v. Jordan, 415 U. S. 651 (1974); Ford Motor Co. v. Department of Treasury, 323 U. S. 459 (1945).

Ill

In No. 75-283, the state officials contest the Court of Appeals’ conclusion that an award of attorneys’ fees in *457this case would under Edelman have only an “ancillary effect” on the state treasury and could therefore be permitted as falling outside the Eleventh Amendment under the doctrine of Ex parte Young, 209 U. S. 123 (1908). 415 U. S., at 667-668. We need not address this question, since, given the express congressional authority for such an award in a case brought under Title VII,12 it follows necessarily from our holding in No. 75-251 that Congress’ exercise of power in this respect is also not barred by the Eleventh Amendment. We therefore affirm the Court of Appeals’ judgment in No. 75-283 on this basis.

The judgment in No. 75-251 is

Reversed.

The judgment in No. 75-283 is

Affirmed.

Mr. Justice Brennan,

concurring in the judgment.

This suit was brought by present and retired employees of the State of Connecticut against the State Treasurer, the State Comptroller, and the Chairman of the State Employees’ Retirement Commission. In that circumstance, Connecticut may not invoke the Eleventh Amendment, since that Amendment bars only federal-court suits against States by citizens of other States. Rather, the question is whether Connecticut may avail itself of the nonconstitutional but ancient doctrine of sovereign immunity as a bar to a claim for damages under Title VII. In my view Connecticut may not assert sovereign immunity for the reason I expressed in dissent in Employees v. Missouri Public Health Dept., 411 U. S. 279, 298 (1973): The States surrendered that immunity, in Hamilton’s words, “in the plan of the Convention” that formed the Union, at least insofar as the States granted Congress *458specifically enumerated powers. See id., at 319 n. 7; Edelman v. Jordan, 415 U. S. 651, 687 (1974) (Brennan, J., dissenting); Parden v. Terminal R. Co., 377 U. S. 184 (1964). Congressional authority to enact the provisions of Title VII at issue in this case is found in the Commerce Clause, Art. I, § 8, cl. 3, and in § 5 of the Fourteenth Amendment, two of the enumerated powers granted Congress in the Constitution. Cf. Oregon v. Mitchell, 400 U. S. 112, 131-134 (1970) (Black, J.); id., at 135-150 (Douglas, J.); id., at 216-217 (Harlan, J.); id., at 236-281 (Brennan, White, and Marshall, JJ.) ; id., at 282-284 (Stewart, J.); Katzenbach v. Morgan, 384 U. S. 641, 651 (1966). I remain of the opinion that “because of its surrender, no immunity exists that can be the subject of a congressional declaration or a voluntary waiver.” Employees v. Missouri Public Health Dept., supra, at 300.

I therefore concur in the judgment of the Court.

Mr. Justice Stevens,

concurring in the judgment.

In my opinion the commerce power is broad enough to support federal legislation regulating the terms and conditions of state employment and, therefore, provides the necessary support for the 1972 Amendments to Title VII, even though Congress expressly relied on § 5 of the Fourteenth Amendment. But I do not believe plaintiffs proved a violation of the Fourteenth Amendment, and because I am not sure that the 1972 Amendments were “needed to secure the guarantees of the Fourteenth Amendment,” see Katzenbach v. Morgan, 384 U. S. 641, 651, I question whether § 5 of that Amendment is an adequate reply to Connecticut’s Eleventh Amendment defense. I believe the defense should be rejected for a different reason.

Even if the Eleventh Amendment does cover a citizen’s *459suit against his own State,1 it does not bar an action against state officers enforcing an invalid statute, Ex parte Young, 209 U. S. 123, 159-160. Since the Connecticut pension law has been held to be invalid, at least in part, Ex parte Young makes it clear that the federal court properly acquired jurisdiction of the proceeding.

The Eleventh Amendment issue presented is whether the court has power to enter a judgment payable immediately out of trust assets which subsequently would be reimbursed from the general revenues of the State. Although I have great difficulty with a construction of the Eleventh Amendment which acknowledges the federal court’s jurisdiction of a case and merely restricts the kind of relief the federal court may grant,2 I must recognize that it has been so construed in Edelman v. Jordan, 415 TJ. S. 651, and that the language of that opinion would seem to cover this case. However, its actual holding appears to be limited to the situation in which the award is payable directly from state funds and “not as a necessary consequence of compliance in the future” with a substantive determination. Id., at 668.

The holding in Edelman does not necessarily require the same result in this case; this award will not be paid directly from the state treasury, but rather from two *460separate and independent pension funds. The fact that the State will have to increase its future payments into the funds as a consequence of this award does not, in my opinion, sufficiently distinguish this case from other cases in which a State may be required to conform its practices to the Federal Constitution and thereby to incur additional expense in the future. Since the rationale of Ex parte Young remains applicable to such cases, and since this case is not squarely covered by the holding in Edelman, I am persuaded that it is proper to reject the Eleventh Amendment defense.

With respect to the fee issue, even if the Eleventh Amendment were applicable, I would place fees in the same category as other litigation costs. Cf. Fairmont Co. v. Minnesota, 275 U. S. 70.

*269Mr. Chief Justice Burger

delivered the opinion of the Court.

We granted certiorari in this case to consider two questions concerning the remedial powers of federal district courts in school desegregation cases, namely, whether a District Court can, as part of a desegregation decree, order compensatory or remedial educational programs for schoolchildren who have been subjected to past acts of de jure segregation, and whether, consistent with the Eleventh Amendment, a federal court can require state officials found responsible for constitutional violations to bear part of the costs of those programs.

I

This case is before the Court for the second time, following our remand, Milliken v. Bradley, 418 U. S. 717 (1974) (Milliken I); it marks the culmination of seven years of litigation over de jure school segregation in the Detroit public school system. For almost six years, the litigation has focused exclusively on the appropriate remedy to correct official acts of racial discrimination committed by both the Detroit School Board and the State of Michigan. No challenge is now made by the State" oFihelocal school board to the prior findings of de jure segregation.1

*270A

In the first stage of the. remedy proceedings, which we reviewed in Milliken I, supra, the District Court, after reviewing several “Detroit-only” desegregation plans, concluded that an interdistrict plan was required to “ 'achieve the greatest degree of actual desegregation ... [so that] no school, grade or classroom [would be] substantially disproportionate to the overall pupil racial composition.’ ” 34511. Supp. 914, 918 (ED Mich. 1972), quoted in Milliken I, supra, at 734. On those premises, the District Court ordered the parties to submit plans for “metropolitan desegregation” and appointed a nine-member panel to formulate a desegregation plan, which would encompass a “desegregation area” consisting of 54 school districts,

. In June 1973, a divided Court of Appeals, sitting en banc, upheld, 484 F. 2d 215 (CA6), the District Court’s determination that a metropolitanwide plan was essential to bring about what the District Court had described as “the greatest degree of actual desegregation . . . .” 345 F. Supp., at 918. We reversed, holding that the order exceeded appropriate limits of federal equitable authority as defined in Swann v. Charlotte-Mecklenburg Board of Education, 402 U. S. 1, 24 (1971), by concluding that “as a matter of substantive constitutional right, [a] particular degree of racial balance” is required, and by subjecting other school districts, uninvolved with and unaffected by any constitutional violations, to the court’s remedial powers. Milliken I, supra. Proceeding from the Swann standard “that the scope of the remedy is determined by the nature and extent of the constitutional violation,” we held that, on the record before us, there was no interdistrict viola*271tion calling for an interdistrict remedy. Because the District Court’s “metropolitan remedy” went beyond the constitutional violation, we remanded the case for further proceedings “leading to prompt formulation of a decree directed to eliminating the segregation found to exist in the Detroit city schools, a remedy which has been delayed since 1970.” 418 U. S., at 753.2

B

Due to the intervening death of Judge Stephen J. Roth, who had presided over the litigation from the outset, the case on remand was reassigned to Judge Robert E. DeMascio. Judge DeMascio promptly ordered respondent Bradley and the Detroit Board to submit desegregation plans limited to the Detroit school system. On April 1, 1975, both parties submitted their proposed plans. Respondent Bradley’s plan was limited solely to pupil reassignment; the proposal called for extensive transportation of students to achieve the plan’s ultimate goal of assuring that every school within the district reflected, within 15 percentage points, the racial ratio of the school district as a whole.3 In contrast to respondent Brad*272ley’s proposal, the Detroit Board’s plan provided for sufficient pupil reassignment to eliminate “racially identifiable white elementary schools,” while ensuring that “every child will spend at least a portion of his education in either a neighborhood elementary school or a neighborhood junior and senior high school.” 402 F. Supp. 1096, 1116 (1975). By eschewing racial ratios for each school, the Board’s plan contemplated transportation of fewer students for shorter distances than respondent Bradley’s proposal.4

In addition to student reassignments, the Board’s plan called for implementation of 13 remedial or compensatory programs, referred to in the record as “educational components.” These compensatory programs, which were proposed in addition to the plan’s provisions for magne/Lschools and vocational high schools, included three of the four components at issue in ’this 'case — in-service training for teachers and administrators, guidance and counseling programs, and revised testing procedures.5 Pursuant to the District Court’s direction, the State Board of Education 6 on April 21, 1975, *273submitted a critique of the Detroit Board’s desegregation plan; in its report, the State Board opined that, although “[i]t is possible that none of the thirteen ‘quality education’ components is essential ... to correct the constitutional violation ...,” 8 of the 13 proposed programs nonetheless deserved special consideration in the desegregation setting. Of particular relevance here, the State Board said:

“Within the context of effectuating a pupil desegregation plan, the in-service training [and] guidance and counseling . . . components appear to deserve special emphasis.” 4 Record, Doc. 591, pp. 38-39.7

After receiving the State Board’s critique,8 the District Court conducted extensive hearings on the two plans over a two-month period. Substantial testimony was adduced with respect to the proposed educational components, including testimony by petitioners’ expert witnesses.9 Based on this *274evidence and on reports of court-appointed experts, the District Court on August 11, 1975, approved, in principle, the Detroit Board’s inclusion of remedial and compensatory educational components in the desegregation plan.10

“We find that the majority of the educational com- / ponents included in the Detroit Board plan are essential / for a school district undergoing desegregation. While it is ¡ true that the delivery of quality desegregated educational | services is the obligation of the school board, nevertheless ’t this court deems it essential to mandate educational com- \ ponents where they are needed to remedy effects of past \ segregation, to assure a successful desegregative effort and \ to minimize the possibility of resegregation.” 402 F. \^upp., at 1118.

The District Court expressly found that the two components of testing and counseling, as then administered in Detroit’s *275schools, were infected with the discriminatory bias of a segregated school system:

“In a segregated setting many techniques deny equal protection to black students, such as discriminatory testing [and] discriminatory counseling . . . Ibid.

The District Court also found that, to make desegregation work, it was necessary to include remedial reading programs and in-service training for teachers and administrators:

“In a system undergoing desegregation, teachers will require orientation and training for desegregation. . . . Additionally, we find that . . . comprehensive reading programs are essential ... to a successful desegregative effort.” Ibid.

Having established these general principles, the District Court formulated several “remedial guidelines” to govern the Detroit Board’s development of a final plan. Declining “to substitute its authority for the authority of elected state and local officials to decide which educational components are beneficial, to the school community,” id., at 1145, the District Judge laid down the following guidelines with respect to each of the four educational components at issue here:

(a) Reading. Concluding that “[t]here is no educational component more directly associated with the process of desegregation than reading,” id., at 1138, the District Court directed the General Superintendent of Detroit’s schools to institute a remedial reading and communications skills program “[t]o eradicate the effects of past discrimination . . . .” Ibid. The content of the required program was not prescribed by the court; rather, formulation and implementation of the program was left to the Superintendent and to a committee to be selected by him.

(b) In-Service Training. The court also directed the Detroit Board to formulate a comprehensive in-service teacher *276training program, an element “essential to a system undergoing desegregation.” Id,., at 1139. In the District Court’s view, an in-service training program for teachers and administrators, to tram professional and instructional personnel to cope with the desegregation process in Detroit, would tend to ensure that all students in a desegregated system would be treated equally by teachers and administrators able, by virtue of special training, to cope with special problems presented by desegregation, and thereby facilitate Detroit’s conversion to a unitary system.

(c) Testing. Because it found, based on record evidence, that Negro children “are especially affected by biased testing procedures,” the District Court determined that, frequently, minority students in Detroit were adversely affected by discriminatory testing procedures. Unless the school system’s tests were administered in a way “free from racial, ethnic and cultural bias,” the District Court concluded that Negro children in Detroit might thereafter be impeded in their educar tional growth. Id., at 1142. Accordingly, the court directed the Detroit Board and the State Department of Education to institute a testing program along the lines proposed by the local school board in its original desegregation plan. Ibid.

(d) Counseling and Career Guidance. Finally, the District Court addressed what expert witnesses had described as psychological pressures on Detroit’s students in a system undergoing desegregation. Counselors were required, the court concluded, both to deal with the numerous problems and tensions arising in the change from Detroit’s dual system, and, more concretely, to counsel students concerning the new vocational and technical school programs available under the plan through the cooperation of state and local officials.11

*277Nine months later, on May 11, 1976, the District' Court A entered its final order. Emphasizing that it had “been careful to order only what is essential for a school district undergoing desegregation,” App. to Pet. for Cert. 117a, the court ordered the Detroit Board and the state defendants to institute comprehensive programs as to the four educational components by the start of the September 1976 school term. The cost of these four programs, the court concluded, was to be equally borne by the Detroit School Board and the State. To / carry out this cost sharing, the court directed the local board to calculate its highest budget allocation in any prior year for/ the several educational programs and, from that base, any/ excess cost attributable to the desegregation plan was to be paid equally by the two groups of defendants responsible for prior constitutional violations, i. e., the Detroit Board and the state defendants.

On appeal, the Court of Appeals for the Sixth Circuit affirmed the District Court’s order concerning the implementation of and cost sharing for the four educational components.12 540 F. 2d 229 (1976). The Court of Appeals ex*278pressly approved the District Court’s findings as to the necessity for these compensatory programs:

“This finding ... is not clearly erroneous, but to the contrary is supported by ample evidence.

“The need for in-service training of the educational staff and development of nondiscriminatory testing is obvious. The former is needed to insure that the teachers and administrators will be able to work effectively in a desegregated environment. The latter is needed to insure that students are not evaluated unequally because of built-in bias in the tests administered in formerly segregated schools.

“We agree with the District Court that the reading and counseling programs are essential to the effort to combat the effects of segregation.

“Without the reading and counseling components, black students might be deprived of the motivation and achievement levels which the desegregation remedy is designed to accomplish.” Id., at 241.

After reviewing the record, the Court of Appeals confirmed that the District Court relied largely on the Detroit School Board in formulating the decree:

“This is not a situation where the District Court ‘appears to have acted solely according to its own notions of good educational policy unrelated to the demands of the Constitution.’ ” Id., at 241-242, quoting Keyes v. School Dist. No. 1, Denver, Colo., 521 F. 2d 465, 483 (CA10 1975), cert. denied, 423 U. S. 1066 (1976).

After upholding the remedial-components portion of the plan, the Court of Appeals went on to affirm the District Court’s allocation of costs between the state and local officials. Analyzing this Court’s decision in Edelman v. Jordan, 415 U. S. 651 (1974), which reaffirmed the rule that the Eleventh *279Amendment bars an ordinary suit for money damages against the State without its consent, the Court of Appeals held:

“[The District Court’s order] imposes no money judgment on the State of Michigan for past de jure segregation practices. Rather, the order is directed toward the State defendants as a part of a prospective plan to comply with a constitutional requirement to eradicate all vestiges of de jure segregation.” 540 F. 2d, at 245. (Emphasis supplied.)

The Court of Appeals remanded the case for further consideration of the three central city regions untouched by the District Court’s pupil reassignment plan. See n. 12, supra.

The state defendants then sought review in this Court, challenging only those portions of the District Court’s comprehensive remedial order dealing with the four educational components and with the State’s obligation to defray the costs of those programs. We granted certiorari, 429 U. S. 958 (1976), and we affirm.

II

This Court has not previously addressed directly the question whether federal courts can order remedial education programs as part of a school desegregation decree.13 However, the general principles governing our resolution of this issue are well settled by the prior decisions of this Court. In the first case concerning federal courts’ remedial powers in eliminating de jure school segregation, the Court laid down the basic rule which governs to this day: “In fashioning and *280effectuating the [desegregation] decrees, the courts will be guided by equitable principles.” Brown v. Board of Education, 349 U. S. 294, 300 (1955) (Brown II).

A

Application of those “equitable principles,” we have held, requires federal courts to focus upon three factors. In the first place, like other equitable remedies, the nature of the desegregation remedy is to be determined by the nature and scope of the constitutional violation. Swann v. Charlotte-Mecklenburg Board of Education, 402 U. S., at 16. The remedy must therefore be related to “the condition alleged to offend the Constitution . . . .” Milliken I, 418 U. S., at 738.14 Second, the decree must indeed be remedial in nature, that is, it must be designed as nearly as possible “to restore the victims of discriminatory conduct to the position they would have occupied in the absence of such conduct.” Id., at 746.15 Third, the federal courts in devising a remedy must *281take into account the interests of state and local authorities in managing their own affairs, consistent with the Constitution. In Brown II the Court squarely held that “[sjchool authorities have the primary responsibility for elucidating, assessing, and solving these problems . . . 349 U. S., at 299. (Emphasis supplied.) If, however, “school authorities fail in their affirmative obligations . . . judicial authority may be invoked.” Swann, supra, at 15. Once invoked, “the scope of a district court’s equitable powers to remedy past wrongs is broad, for breadth and flexibility are inherent in equitable remedies.” Ibid.

B

In challenging the order before us, petitioners do not specifically question that the District Court’s mandated programs are designed, as nearly as practicable, to restore the schoolchildren of Detroit to the position they would have enjoyed absent constitutional violations by state and local officials. And, petitioners do not contend, nor could they, that the prerogatives of the Detroit School Board have been abrogated by the decree, since of course the Detroit School Board itself proposed incorporation of these programs in the first place. Petitioners’ sole contention is that, under Swann, the District Court’s order exceeds the scope of the constitutional violation. Invoking our holding in Milliken I, petitioners claim that, since the constitutional violation found by the District Court was the unlawful segregation of students on the basis of race, the, court’s decree must be-limited to remedying unlawful pupil assignments. This contention misconceives the principle petitioners seek to invoke, and we reject their argument.

The well-settled principle that the nature and scope of *282the remedy axe to be determined by the violation means simply that federal-court decrees must directly address and relate to the constitutional violation itself. Because of this inherent limitation upon federal judicial authority, federal-court decrees exceed appropriate limits if they are aimed at eliminating a condition that does not violate the Constitution or does not flow from such a violation, see Pasadena Bd. of Education v. Spangler, 427 U. S. 424 (1976), or if they are imposed upon governmental units that were neither involved in nor affected by the constitutional violation, as in Milliken I, supra. Hills v. Gautreaux, 425 U. S. 284, 292-296 (1976). But where, as here, a constitutional violation has been found, the remedy does not “exceed” the violation if the remedy is tailored to cure the “ ‘condition that offends the Constitution.' ” Milliken I, supra, at 738. (Emphasis supplied.)

The “condition” offending the Constitution is Detroit's de jure segregated school system, which was so pervasively and persistently segregated that the District Court found that the need for the educational components flowed directly from constitutional violations by both state and local officials. These specific educational remedies, although normally left to the discretion of the elected school board and professional educators, were deemed necessary to restore the victims of discriminatory conduct to the position they would have enjoyed in terms of education had these four components been provided in a nondiscriminatory manner in a school system free from pervasive de jure racial segregation.

In the first case invalidating a de jure system, a unanimous Court, speaking through Mr. Chief Justice Warren, held in Brown v. Board of Education, 347 U. S. 483, 495 (1954) (Brown /): “Separate educational facilities are inherently unequal.” And in United States v. Montgomery County Bd. of Educ., 395 U. S. 225 (1969), the Court concerned itself not with pupil assignment, but with the desegregation of faculty and staff as part of the process of dismantling a dual *283system. In doing so, the Court, there speaking through Mr. Justice Black, focused on the reason for judicial concerns going beyond pupil assignment: “The dispute . . . deals with faculty and staff desegregation, a goal that we have recognized to be an important aspect of the basic task of achieving a public school system wholly free from racial discrimination.” Id., at 231-232. (Emphasis supplied.)

Montgomery County therefore stands firmly for the proposition that matters other than pupil assignment must on occasion be addressed by federal courts to eliminate the effects of prior segregation. Similarly, in Swann we reaffirmed the principle laid down in Green v. County School Bd., 391 U. S. 430 (1968), that “existing policy and practice with regard to faculty, staff, transportation, extracurricular activities, and facilities were among the most important indicia of a segregated system.” 402 U. S., at 18. In a word, discriminatory student assignment policies can themselves manifest and breed other inequalities built into a dual system founded on racial discrimination. Federal courts need not, and cannot, close their eyes to inequalities, shown by the record, which flow from a longstanding segregated system.

C

In light of the mandate of Brown I and Brown II, federal courts have, over the years, often required the inclusion of remedial programs in desegregation plans to overcome the inequalities inherent in dual school systems. In 1966, for example, the District Court for the District of South Carolina directed the inclusion of remedial courses to overcome the effects of a segregated system:

“Because the weaknesses of a dual school system may have already affected many children, the court would be remiss in its duty if any desegregation plan were approved which did not provide for remedial education courses. They shall be included in the plan.” Miller v. School *284District 2, Clarendon County, S. C., 256 F. Supp. 370, 377 (1966).

In 1967, the Court of Appeals for the Fifth Circuit, then engaged in overseeing the desegregation of numerous school districts in the South, laid down the following requirement in an en banc decision:

“The defendants shall-provide remedial education programs which permit students attending or who have previously attended segregated schools to overcome past inadequacies in their education.” United States v. Jefferson County Board of Education, 380 F. 2d 385, 394, cert. denied, 389 U. S. 840 (1967). (Emphasis supplied.)

See also Stell v. Board of Public Education of Savannah, 387 F. 2d 486, 492, 496-497 (CA5 1967); Hill v. Lafourche Parish School Board, 291 F. Supp. 819, 823 (ED La. 1967); Redman v. Terrebonne Parish School Board, 293 F. Supp. 376, 379 (ED La. 1967); Lee v. Macon County Board of Education, 267 F. Supp. 458, 489 (MD Ala. 1967); Graves v. Walton County Board of Education, 300 F. Supp. 188, 200 (MD Ga. 1968), aff’d, 410 F. 2d 1153 (CA5 1969). Two years later, the Fifth Circuit again adhered to the rule that district courts could properly seek to overcome the built-in inadequacies of a segregated educational system:

“The trial court concluded that the school board must establish remedial programs to assist students who previously attended all-Negro schools when those students transfer to formerly all-white schools .... The remedial programs . . . are an integral part of a program for compensatory education to be provided Negro students who have long been disadvantaged by the inequities and discrimination inherent in the dual school system. The requirement that the School Board institute remedial programs so far as they are feasible is a proper exercise of the court’s discretion.” Plaquemines Parish School Bd. v. *285United States, 415 F. 2d 817, 831 (1969). (Emphasis supplied.)

In the same year the United States District Court for the Eastern District of Louisiana required school authorities to come forward with a remedial educational program as part of a desegregation plan. “ 'The defendants shall provide remedial education programs which permit students . . . who have previously attended all-Negro schools to overcome past inadequacies in their education.’ ” Smith v. St. Tammany Parish School Board, 302 F. Supp. 106, 110 (1969), aff’d, 448 F. 2d 414 (CA5 1971). See also Moore v. Tangipahoa Parish School Board, 304 F. Supp. 244, 253 (ED La. 1969); Moses v. Washington Parish School Board, 302 F. Supp. 362, 367 (ED La. 1969).

In the 1970’s, the pattern has been essentially the same. The Fifth Circuit has, when the fact situation warranted, continued to call for remedial education programs in desegregation plans. E. g., United States v. Texas, 447 F. 2d 441, 448 (1971), stay denied sub nom. Edgar v. United States, 404 U. S. 1206 (1971) (Black, J., in chambers). To that end, the approved plan in United States v. Texas required:

''[Curriculum offerings and programs shall include specific educational programs designed to compensate minority group children for unequal educational opportunities resulting from past or present racial and ethnic isolation . . . .” 447 F. 2d, at 448.16

See also George v. O’Kelly, 448 F. 2d 148, 150 (CA5 1971). And, as school desegregation litigation emerged in other *286regions of the country, federal courts have likewise looked in part to remedial programs, when the record supported an order to that effect. See, e. g., Morgan v. Kerrigan, 401 F. Supp. 216, 235 (Mass. 1975), aff’d, 530 F. 2d 401 (CA1), cert. denied sub nom. White v. Morgan, 426 U. S. 935 (1976); Hart v. Community School Board of Brooklyn, 383 F. Supp. 699, 757 (EDNY 1974), aff’d, 512 F. 2d 37 (CA2 1975); cf. Booker v. Special School Dist. 1, Minneapolis, Minn., 351 F. Supp. 799 (Minn. 1972).17

Finally, in addition to other remedial programs, which could, if circumstances warranted, include programs to remedy deficiencies, particularly in reading and communications skills, federal courts have expressly ordered special in-service training for teachers, see, e. g., United States v. Missouri, 523 F. 2d 885, 887 (CA8 1975); Smith v. St. Tammany Parish School Board, supra, at 110; Moore v. Tangipahoa Parish School Board, supra, at 253, and have altered or even suspended testing programs employed by school systems undergoing desegregation. See, e. g., Singleton v. Jackson Municipal Separate School Dist., 419 F. 2d 1211, 1219 (CA5 1969), cert. denied, 396 U. S. 1032 (1970); Lemon v. Bossier Parish School Board, 444 F. 2d 1400, 1401 (CA5 1971); Arvizu v. Waco Independent School Dist., 373 F. Supp. 1264 (WD Tex. 1973), rev’d in part on other issues, 495 F. 2d 499 (CA5 1974).

Our reference to these cases is not to be taken as necessarily approving holdings not reviewed by this Court. However, they demonstrate that the District Court in the case now *287before us did not break new ground in approving the School Board’s proposed plan. Quite the contrary, acting on abundant evidence in this record, the District Court approved a remedial plan going beyond mere pupil assignments, as expressly approved by Swann and Montgomery County. In so doing, the District Court was adopting specific programs proposed by local school authorities, who must be presumed to be familiar with the problems and the needs of a system undergoing desegregation.18

We do not, of course, imply that the order here is a blueprint for other cases. That cannot be; in school desegregation cases, “[t]here is no universal answer to complex problems . . . ; there is obviously no one plan that will do the job in every case.” Green, 391 U. S., at 439. On this record, however, we are bound to conclude that the decree before us was aptly tailored to remedy the consequences of the constitutional violation. Children who have been thus educationally and culturally set apart from the larger community will inevitably acquire habits of speech, conduct, and attitudes reflecting their cultural isolation. They are likely to acquire speech habits, for example, which vary from the environment in which they must ultimately function and compete, if they are to enter and be a part of that community. This is not peculiar to race; in this setting, it can affect any children who, as a group, are isolated by force of law from the mainstream. Cf. Lau v. Nichols, 414 U. S. 563 (1974).

Pupil assignment alone does not automatically remedy the impact of previous, unlawful educational isolation; the consequences linger and can be dealt with only by inde*288pendent measures. In short, speech habits acquired in a segregated system do not vanish simply by moving the child to a desegregated school. The root condition shown by this record must be treated directly by special training at the hands of teachers prepared for that task. This is what the District Judge in the case drew from the record before him as to the consequences of Detroit’s de jure system, and we cannot conclude that the remedies decreed exceeded the scope of the violations found.

Nor do we find any other reason to believe that the broad and flexible equity powers of the court were abused in this case. The established role of local school authorities was maintained inviolate, and the remedy is indeed remedial, The order does not punish anyone, nor does it impair or jeopardize the educational system in Detroit.19 The District Court, in short, was true to the principle laid down in Brown II:

“In fashioning and effectuating the decrees, the courts will be guided by equitable principles. Traditionally, equity has been characterized by a practical flexibility in shaping its remedies and by a facility for adjusting and reconciling public and private needs. These cases call for the exercise of these traditional attributes of equity power.” 349 U. S., at 300 (footnotes omitted).

Ill

Petitioners also contend that the District Court’s order, even if otherwise proper, violates the Eleventh Amendment. In their view, the requirement that the state defendants pay one-half the additional costs attributable to the four educa*289tional components is, “in practical effect, indistinguishable from an award of money damages against the state based upon the asserted prior misconduct of state officials.” Brief for Petitioners 34. Arguing from this premise, petitioners conclude that the “award” in this case is barred under this Court’s holding in Edelman v. Jordan, 415 U. S. 651 (1974).

Edelman involved a suit for money damages against the State, as well as for prospective injunctive relief.20 The suit was brought by an individual who claimed that Illinois officials had improperly withheld disability benefit payments from him and from the members of his class. Applying traditional Eleventh Amendment principles, we held that the suit was barred to the extent the suit sought “the award of an accrued monetary liability . . .” which represented “retroactive payments.” Id., at 663-664. (Emphasis supplied.) Conversely, the Court held that the suit was proper to the extent it sought “payment of state funds ... as a necessary consequence of compliance in the future with a substantive federal-question determination . . . .” Id., at 668. (Emphasis supplied.)

The decree to share the future costs of educational components in this case fits squarely within the prospective-compliance exception reaffirmed by Edelman. That exception, which had its genesis in Ex parte Young, 209 U. S. 123 (1908), permits federal courts to enjoin state officials to conform their conduct to requirements of federal law, notwithstanding a direct and substantial impact on the state treasury. 415 U. S., at 667. The order challenged here does no more than that. The decree requires state officials, held responsible for unconstitutional conduct, in findings which are not challenged, to eliminate a de jure segregated school system. More precisely, the burden of state officials is that set forth *290in Swann — -to take the necessary steps “to eliminate from the public schools all vestiges of state-imposed segregation.” 402 U. S., at 15. The educational components, which the District Court ordered into effect prospectively, are plainly designed to wipe out continuing conditions of inequality produced by the inherently unequal dual school system long maintained by Detroit.21

These programs were not, and as a practical matter could not be, intended to wipe the slate clean by one bold stroke, as could a retroactive award of money in Edelman.22 Rather, by the nature of the antecedent violation, which on this record caused significant deficiencies in communications skills— reading and speaking — the victims of Detroit’s de jure segregated system will continue to experience the effects of segregation until such future time as the remedial programs can help dissipate the continuing effects of past misconduct. Reading and speech deficiencies cannot be eliminated by judicial fiat; they will require time, patience, and the skills of specially trained teachers. That the programs are also “compensatory” in nature does not change the fact that they are part of a plan that operates prospectively to bring about the delayed benefits of a unitary school system. We therefore hold that such prospective relief is not barred by the Eleventh Amendment.23

*291Finally, there is no merit to petitioners’ claims that the relief ordered here violates the Tenth Amendment and general principles of federalism. The Tenth Amendment’s reservation of nondelegated powers to the States is not implicated by a federal-court judgment enforcing the express prohibitions of unlawful state conduct enacted by the Fourteenth Amendment. Cf. Fitzpatrick v. Bitzer, 427 U. S. 445 (1976). Nor are principles of federalism abrogated by the decree. The District Court has neither attempted to restructure local governmental entities nor to mandate a particular method or structure of state or local financing. Cf. San Antonio School Dist. v. Rodriguez, 411 U. S. 1 (1973). The District Court has, rather, properly enforced the guarantees of the Fourteenth Amendment consistent with our prior holdings, and in a manner that does not jeopardize the integrity of the structure or functions of state and local government.

The judgment of the Court of Appeals is therefore

Affirmed.

Mr. Justice Marshall,

concurring.

I wholeheartedly join The Chief Justice’s opinion for the Court. My Brother Powell’s opinion prompts these additional comments.

What is, to me, most tragic about this case is that in all relevant respects it is in no way unique. That a northern school board has been found guilty of intentionally discriminatory acts is, unfortunately, not unusual. That the academic development of black children has been impaired by this wrongdoing is to be expected. And, therefore, that a program *292of remediation is necessary to supplement the primary remedy of pupil reassignment is inevitable.

It is of course true, as Mr. Justice Powell notes, that the Detroit School Board has belatedly recognized its responsibility for the injuries that Negroes have suffered, and has joined in the effort to remedy them.- He may be right— although I hope not — that this makes the case “wholly different from any prior case,” post, this page. But I think it worth noting that the legal issues would be no different if the Detroit School Board came to this Court on the other side. The question before us still would be the one posed by the State: Is the remedy tailored to fit-the scope of the violation? And, as The Chief Justice convincingly demonstrates, that question would have to be answered in the affirmative in light of the findings of the District Court, supported by abundant evidence. Cf. Dayton Board of Education v. Brinkman, post, at 414.

Mr. Justice Powell,

concurring in the judgment.

The Court's opinion addresses this case as if it were conventional desegregation litigation. The wide-ranging opinion reiterates the familiar general principles drawn from the line of precedents commencing with Brown v. Board of Education, 347 U. S. 483 (1954), and including today’s decision in Dayton Board of Education v. Brinkman, post, p. 406. One has to read the opinion closely to understand that the case, as it finally reaches us, is wholly different from any prior case. I -write to emphasize its uniqueness, and the consequent limited precedential effect of much of the Court’s opinion.

Normally, the plaintiffs in this type of litigation are students, parents, and supporting organizations that desire to desegregate a school system alleged to be the product, in whole or in part, of de jure segregative action by the public school authorities. The principal defendant is usually the *293local board of education or school board. Occasionally the state board of education and state officials are joined as defendants. This protracted litigation commenced in 1970 in this conventional mold. In the intervening years, however, the posture of the litigation has changed so drastically as to leave it largely a friendly suit between the plaintiffs (respondents Bradley et al.) and the original principal defendant, the Detroit School Board. These parties, antagonistic for years, have now joined forces apparently for the purpose of extracting funds from the state treasury. As between the original principal parties — the plaintiffs and the Detroit School Board — no case or controversy remains on the issues now before us. The Board enthusiastically supports the entire desegregation decree even though the decree intrudes deeply on the Board’s own decisionmaking powers. Indeed, the present School Board proposed most of the educational components included in the District Court’s decree. The plaintiffs originally favored a desegregation plan that would have required more extensive transportation of pupils, and they did not initially propose or endorse the educational components. In this Court, however, the plaintiffs also support the decree of the District Court as affirmed by the Court of Appeals.1

Thus the only complaining party is the State of Michigan (acting through state officials) and its basic complaint concerns money, not desegregation. It has been ordered to pay about $5,800,000 to the Detroit School Board. This is one-half the estimated “excess cost” of 4 of the 11 educational components *294included in the desegregation decree: remedial reading, in-service training of teachers, testing, and counseling.2 The State, understandably anxious to preserve the state budget from federal-court control or interference, now contests the decree on two grounds.

*295First, it is argued that the order to pay state funds violates the Eleventh Amendment and principles of federalism. Ordinarily a federal court’s order that a State pay unappropriated funds to a locality would raise the gravest constitutional issues. See generally San Antonio School Dist. v. Rodriguez, 411 U. S. 1, 40-42 (1973); National League of Cities v. Usery, 426 U. S. 833 (1976). But here, in a finding no longer subject to review, the State has been adjudged a participant in the constitutional violations, and the State therefore may be ordered to participate prospectively in a remedy otherwise appropriate.

The State’s second argument is one that normally would be advanced vigorously by the school board. Relying on the established principle that the scope of the remedy in a desegregation case is determined and limited by the extent of the identified constitutional violations, Dayton Board of Education, post, at 419-420; Hills v. Gautreaux, 425 U. S. 284, 293-294 (1976); Milliken v. Bradley, 418 U. S. 717, 744 (1974); Austin Independent School Dist. v. United States, 429 U. S. 990, 991 (1976) (Powell, J., concurring), the State argues that the District Court erred in ordering the system-wide expansion of the four educational components mentioned above. It contends that there has been no finding of a constitutional violation with respect to the past operation of any of these programs, and it insists that without more specifically focused findings of this sort, the decree exceeded the court’s powers.

This argument is by no means a frivolous one. But the context in which it is presented is so unusual that it would be appropriate to dismiss the writ as improvidently granted. The argument is advanced by the State and not by the party primarily concerned. The educational programs at issue are standard and widely approved in public education. The State Board normally would be enthusiastic over enhancement of these programs so long as the local school board could *296fund, them without requiring financial aid from the State. It is equally evident that the State probably would resist a federal-court order requiring it to pay unappropriated state funds to the local school board regardless of whether violations by the local board justified the remedy. The State’s interest in protecting its own budget — limited by legislative appropriations — is a genuine one. But it is not an interest that is related, except fortuitously, to a claim that the desegregation remedy may have exceeded the extent of the violations.

The State’s reliance on the remedy issue contains a further weakness, emphasizing the unusual character of this case. There is no indication that the State objected — certainly, it does not object here — to the inclusion in the District Court’s decree of the seven other educational components. See n. 2, supra. Indeed, the State expressly agreed to one of the most expensive components, the establishment of vocational education centers, in a stipulation obligating it to share the cost of construction equally with the Detroit Board. See App. to Pet. for Cert. 139a-144a. Furthermore, the District Court’s decree largely embodies the original recommendation of the Detroit Board. Since local school boards “have the primary responsibility for elucidating, assessing, and solving [the] problems” generated by “[f]ull implementation of . . . constitutional principles” in the local setting, Brown v. Board of Education, 349 U. S. 294, 299 (1955), the State’s limited challenge here is particularly lacking in force.

Moreover, the District Court was faced with a school district in exceptional disarray. It found the structure of the Detroit school system “.chaotic and incapable of effective administration.” App. to Pet. for Cert. 124a. The “general superintendent has little direct authority.” Ibid. Each of the eight regional boards may be preoccupied with “dis-tribut[ing] local board patronage.” Id., at 125a. The “local boards have diverted resources that would otherwise have been *297available for educational purposes to build new offices and other facilities to house this administrative overload.” Ibid. The District Court continued:

“In addition to the administrative chaos, we know of no other school system that is so enmeshed in politics. . . .

“. . . Rather than devoting themselves to the educational system and the desegregative process, board members are busily engaged in politics not only to assure their own re- election but also to defeat others with whom they disagree.” Id., at 125a-126a (footnote omitted).

Referring again to the “political paralysis” and “inefficient bureaucracy” of the system, the court also noted — discouragingly — that the election then approaching “may well [result in] a board of education consisting of members possessing no experience in education.” Id., at 126a. In this quite remarkable situation, it is perhaps not surprising that the District Court virtually assumed the role of school superintendent and school board.3

*298Given the foregoing unique circumstances, it seems to me that the proper disposition of this case is to dismiss the writ of certiorari as improvidently granted. But as the Court has chosen to decide the case here, I join in the judgment as a result less likely to- prolong the disruption of education in Detroit than a reversal or remand. Despite wide-ranging dicta in the Court’s opinion, the only issue decided is that the District Court’s findings as to specific constitutional violations justified the four remedial educational components included in the desegregation decree. In my view,, it is at least arguable that the findings in this respect were too generalized to meet-the standards prescribed by this Court. See Dayton Board of Education, post, p. 406. But the majority views the record as justifying the conclusion that “the need for educational components flowed directly from constitutional violations by both state and local officials.” Ante, at 282.4 On that view of the record, our settled doctrine requiring that the remedy be carefully tailored to fit identified constitutional violations is reaffirmed by today’s result. I therefore concur in the judgment.

Mr. Justice Stevens

delivered the opinion of the Court.†

After finding that conditions in the Arkansas penal system constituted cruel and unusual punishment, the District Court entered a series of detailed remedial orders. On appeal to the United States Court of Appeals for the Eighth Circuit, petitioners1 challenged two aspects of that relief: (1) an order placing a maximum limit of 30 days on confinement in punitive isolation; and (2) an award of attorney’s fees to be paid out of Department of Correction funds. The Court of *681Appeals affirmed and assessed an additional attorney’s fee to cover services on appeal. 548 F. 2d 740 (1977). We granted certiorari, 434 U. S. 901, and now affirm.

This litigation began in 1969; it is a sequel to two earlier cases holding that conditions in the Arkansas prison system violated the Eighth and Fourteenth Amendments.2 Only a brief summary of the facts is necessary to explain the basis for the remedial orders.

The routine conditions that the ordinary Arkansas convict had to endure were characterized by the District Court as “a dark and evil world completely alien to the free world.” Holt v. Sarver, 309 F. Supp. 362, 381 (ED Ark. 1970) (Holt II). That characterization was amply supported by the evidence.3 *682The punishments for misconduct not serious enough to result in punitive isolation were cruel,4 unusual,5 and unpredictable.6 It is the discipline known as “punitive isolation” that is most relevant for present purposes.

Confinement in punitive isolation was for an indeterminate period of time. An average of 4, and sometimes as many as 10 or 11, prisoners were crowded into windowless 8'xlO' cells containing no furniture other than a source of water and a toilet that could only be flushed from outside the cell. Holt v. Sarver, 300 F. Supp. 825, 831-832 (ED Ark. 1969) (Holt I). At night the prisoners were given mattresses to spread on the floor. Although some prisoners suffered from infectious diseases such as hepatitis and venereal disease, mattresses were removed and jumbled together each morning, *683then returned to the cells at random in the evening. Id,, at 832. Prisoners in isolation received fewer than 1,000 calories a day;7 their meals consisted primarily of 4-inch squares of “grue,” a substance created by mashing meat, potatoes, oleo, syrup, vegetables, eggs, and seasoning into a paste and baking the mixture in a pan. Ibid.

After finding the conditions of confinement unconstitutional, the District Court did not immediately impose a detailed remedy of its own. Instead, it directed the Department of Correction to “make a substantial start” on improving conditions and to file reports on its progress. Holt I, supra, at 833-834. When the Department’s progress proved unsatisfactory, a second hearing was held. The District Court found some improvements, but concluded that prison conditions remained unconstitutional. Holt II, 309 F. Supp., at 383. Again the court offered prison administrators an opportunity to devise a plan of their own for remedying the constitutional violations, but this time the court issued guidelines, identifying four areas of change that would cure the worst evils: improving conditions in the isolation cells, increasing inmate safety, eliminating the barracks sleeping arrangements, and putting an end to the trusty system. Id., at 385. The Department was ordered to move as rapidly as funds became available. Ibid.

After this order was affirmed on appeal, Holt v. Sarver, 442 F. 2d 304 (CA8 1971), more hearings were held in 1972 and 1973 to review the Department’s progress. Finding substantial improvements, the District Court concluded that continuing supervision was no longer necessary. The court held, *684however, that its prior decrees would remain in effect and noted that sanctions, as well as an award of costs and attorney’s fees, would be imposed if violations occurred. Holt v. Hutto, 363 F. Supp. 194, 217 (ED Ark. 1973) (Holt HI).

The Court of Appeals reversed the District Court’s decision to withdraw its supervisory jurisdiction, Finney v. Arkansas Board of Correction, 505 F. 2d 194 (CA8 1974), and the District Court held a fourth set of hearings. 410 F. Supp. 251 (ED Ark. 1976). It found that, in some respects, conditions had seriously deteriorated since 1973, when the court had withdrawn its supervisory jurisdiction. Cummins Farm, which the court had condemned as overcrowded in 1970 because it housed 1,000 inmates, now had a population of about 1,500. Id., at 254-255. The situation in the punitive isolation cells was particularly disturbing. The court concluded that either it had misjudged conditions in these cells in 1973 or conditions had become much worse since then. Id., at 275. There were twice as many prisoners as beds in some cells. And because inmates in punitive isolation are often violently antisocial, overcrowding led to persecution of the weaker prisoners. The “grue” diet was still in use, and practically all inmates were losing weight on it. The cells had been vandalized to a “very substantial” extent. Id., at 276. Because of their inadequate numbers, guards assigned to the punitive isolation cells frequently resorted to physical violence, using nightsticks and Mace in their efforts to maintain order. Prisoners were sometimes left in isolation for months, their release depending on “their attitudes as appraised by prison personnel.” Id., at 275.

The court concluded that the constitutional violations identified earlier had not been cured. It entered an order that placed limits on the number of men that could be confined in one cell, required that each have a bunk, discontinued the “grue” diet, and set 30 days as the maximum isolation sentence. The District Court gave detailed consideration to *685the matter of fees and expenses, made an express finding that petitioners had acted in bad faith, and awarded counsel “a fee of $20,000.00 to be paid out of Department of Correction funds.” Id,., at 285. The Court of Appeals affirmed and assessed an additional $2,500 to cover fees and expenses on appeal. 548 F. 2d, at 743.

I

The Eighth Amendment’s ban on inflicting cruel and unusual punishments, made applicable to the States by the Fourteenth Amendment, “proscribe[s] more than physically barbarous punishments.” Estelle v. Gamble, 429 U. S. 97, 102. It prohibits penalties that are grossly disproportionate to the offense, Weems v. United States, 217 U. S. 349, 367, as well as those that transgress today’s “ ‘broad and idealistic concepts of dignity, civilized standards, humanity, and decency.’ ” Estelle v. Gamble, supra, at 102, quoting Jackson v. Bishop, 404 F. 2d 571, 579 (CA8 1968). Confinement in a prison or in an isolation cell is a form of punishment subject to scrutiny under Eighth Amendment standards. Petitioners do not challenge this proposition; nor do they disagree with the District Court’s original conclusion that conditions in Arkansas’ prisons, including its punitive isolation cells, constituted cruel and unusual punishment. Rather, petitioners single out that portion of the District Court’s most recent order that forbids the Department to sentence inmates to more than 30 days in punitive isolation. Petitioners assume that the District Court held that indeterminate sentences to punitive isolation always constitute cruel and unusual punishment. This assumption misreads the District Court’s holding.

Read in its entirety, the District Court’s opinion makes it abundantly clear that the length of isolation sentences was not considered in a vacuum. In the court’s words, punitive isolation “is not necessarily unconstitutional, but it may be, depending on the duration of the confinement and the con*686ditions thereof.” 410 F. Supp., at 275.8 It is perfectly obvious that every decision to remove a particular inmate from the general prison population for an indeterminate period could not be characterized as cruel and unusual. If new conditions of confinement are not materially different from those affecting other prisoners, a transfer for the duration of a prisoner’s sentence might be completely unobjectionable and well within the authority of the prison administrator. Cf. Meachum v. Fano, 427 U. S. 215. It is equally plain, however, that the length of confinement cannot be ignored in deciding whether the confinement meets constitutional standards. A filthy, overcrowded cell and a diet of *687“grue” might be tolerable for a few days and intolerably cruel for weeks or months.

The question before the trial court was whether past constitutional violations had been remedied. The court was entitled to consider the severity of those violations in assessing the constitutionality of conditions in the isolation cells. The court took note of the inmates’ diet, the continued overcrowding, the rampant violence, the vandalized cells, and the “lack of professionalism and good judgment on the part of maximum security personnel.” 410 F. Supp., at 277 and 278. The length of time each inmate spent in isolation was simply one consideration among many. We find no error in the court’s conclusion that, taken as a whole, conditions in the isolation cells continued to violate the prohibition against cruel and unusual punishment.

In fashioning a remedy, the District Court had ample authority to go beyond earlier orders and to address each element contributing to the violation. The District Court had given the Department repeated opportunities to remedy the cruel and unusual conditions in the isolation cells. If petitioners had fully complied with the court’s earlier orders, the present time limit might well have been unnecessary. But taking the long and unhappy history of the litigation into account, the court was justified in entering a comprehensive order to insure against the risk of inadequate compliance.9

*688The order is supported by the interdependence of the conditions producing the violation. The vandalized cells and the atmosphere of violence were attributable, in part, to overcrowding and to deep-seated enmities growing out of months of constant daily friction.10 The 30-day limit will help to correct these conditions.11 Moreover, the limit presents little danger of interference with prison administration, for the Commissioner of Correction himself stated that prisoners should not ordinarily be held in punitive isolation for more than 14 days. Id., at 278. Finally, the exercise of discretion in this case is entitled to special deference because of the trial judge’s years of experience with the problem at hand and his recognition of the limits on a federal court’s authority in a case of this kind.12 Like the Court of Appeals, we find no error in the inclusion of a 30-day limitation on sentences to punitive isolation as a part of the District Court’s comprehensive remedy.

*689II

The Attorney General of Arkansas, whose office has represented petitioners throughout this litigation, contends that any award of fees is prohibited by the Eleventh Amendment. He also argues that the Court of Appeals incorrectly held that fees were authorized by the Civil Rights Attorney’s Fees Awards Act of 1976. We hold that the District Court’s award is adequately supported by its finding of bad faith and that the Act supports the additional award by the Court of Appeals.

A. The District Court Award

Although the Attorney General argues that the finding of bad faith does not overcome the State’s Eleventh Amendment protection, he does not question the accuracy of the finding made by the District Court and approved by the Court of Appeals.13 Nor does he question the settled rule that a losing litigant’s bad faith may justify an allowance of fees to the prevailing party.14 He merely argues that the order requir*690ing that the fees be paid from public funds violates the Eleventh Amendment.

In the landmark decision in Ex parte Young, 209 U. S. 123, the Court held that, although prohibited from giving, orders directly to a State, federal courts could enjoin state officials in their official capacities. And in Edelman v. Jordan, 415 U. S. 651, when the Court held that the Amendment grants the States an immunity from retroactive monetary relief, it reaffirmed the principle that state officers are not immune from prospective injunctive relief. Aware that the difference between retroactive and prospective relief “will not in many instances be that between day and night,” id., at 667, the Court emphasized in Edélman that the distinction did not immunize the States from their obligation to obey costly federal-court orders. The cost of compliance is “ancillary” to the prospective order enforcing federal law. Id., at 668.15 The line between retroactive and prospective relief cannot be so rigid that it defeats the effective enforcement of prospective relief.

The present case requires application of that principle. In exercising their prospective powers under Ex parte Young and Edelman v. Jordan, federal courts are not reduced to issuing injunctions against state officers and hoping'for compliance. Once issued, an injunction may be enforced. Many of the court’s most effective enforcement weapons involve financial penalties. A criminal contempt prosecution for “resistance to [the court’s] lawful .’ . . order” may result in a jail term or a fine. 18 U. S. C. § .401 (1976 ed.). Civil contempt proceedings may yield a conditional jail term or fine. United States v. *691Mine Workers, 330 U. S. 258, 305. Civil contempt may also be punished by a remedial fine, which compensates the party who won the injunction for the effects of his opponent’s noncompliance. Id., at 304; Gompers v. Bucks Stove & Range Co., 221 U. S. 418. If a state agency refuses to adhere to a court order, a financial penalty may be the most effective means of insuring compliance. The principles of federalism that inform Eleventh Amendment doctrine surely do not require federal courts to enforce their decrees only by sending high state officials to jail.16 The less intrusive power to impose a fine is properly treated as ancillary to the federal court’s power to impose injunctive relief.

In this case, the award of attorney’s fees for bad faith served the same purpose as a remedial fine imposed for civil contempt. It vindicated the District Court’s authority over a recalcitrant litigant. Compensation was not the sole motive for the award; in setting the amount of the fee, the court said that it would “make no effort to adequately compensate counsel for the work that they have done or for the time that they have spent on the case.” 410 F. Supp., at 285. The court did allow a “substantial” fee, however, because “the allowance thereof may incline the Department to act in such a manner that further protracted litigation about the prisons will not be necessary.” Ibid.17 We see no reason to distin*692guish this award from any other penalty imposed to enforce a prospective injunction.18 Hence the substantive protections of the Eleventh Amendment do not prevent an award of attorney’s fees against the Department’s officers in their official capacities.

Instead of assessing the award against the defendants in their official capacities, the District Court directed that the fees are “to be paid out of Department of Correction funds.” Ibid. Although the Attorney General objects to the form of the order,19 no useful purpose would be served by requiring that it be recast in different language. We have previously approved directives that were comparable in their actual impact on the State without pausing to attach significance to the language used by the District Court.20 Even if it might have *693been better form to omit the reference to the Department of Correction, the use of that language is surely not reversible error.

B. The Court of Appeals Award

Petitioners, as the losing litigants in the Court of Appeals, were ordered to pay an additional $2,500 to counsel for the prevailing parties “for their services on this appeal.” 548 F. 2d, at 743. The order does not expressly direct the Department of Correction to pay the award, but since petitioners are sued in their official capacities, and since they are represented by the Attorney General, it is obvious that the award will be paid with state funds. It is also clear that this order is not supported by any finding of bad faith. It is founded instead on the provisions of the Civil Rights Attorney’s Pees Awards Act of 1976. Pub.. L. No. 94-559, 90 Stat. 2641, 42 U. S. C. § 1988 (1976 ed.). The Act declares that, in suits under 42 U. S. C. § 1983 and certain other statutes, federal courts may award prevailing parties reasonable attorney’s fees “as part of the costs.”'21

As this Court made clear in Fitzpatrick v. Bitzer, 427 U. S. 445, Congress has plenary power to set aside the States’ immunity from retroactive relief in order to enforce the Fourteenth Amendment. When it passed the Act, Congress undoubtedly intended to exercise that power and to authorize fee awards *694payable by the States when their officials are sued in their official capacities. The Act itself could not be broader. It applies to “any” action brought to enforce certain civil rights laws. It contains no hint of an exception for States defending injunction actions; indeed, the Act primarily applies to laws passed specifically to restrain state action. See, e. g., 42 U. S. C. 1 1983.

The legislative history is equally plain: “[I]t is intended that the attorneys’ fees, like other items of costs, will be collected either directly from the official, in his official capacity, from funds of his agency or under his control, or from the State or local government (whether or not the agency or government is a named party).” S. Rep. No. 94^1011, p. 5 (1976) (footnotes omitted). The House Report is in accord: “The greater resources available to governments provide an ample base from which fees can be awarded to the prevailing plaintiff in suits against governmental officials or entities.” H. R. Rep. No. 94U1558, p. 7 (1976). The Report adds in a footnote that: “Of course, the 11th Amendment is not a bar to the awarding of counsel fees against state governments. Fitzpatrick v. Bitzer.” Id., at 7 n. 14. Congress’ intent was expressed in deeds as well as words.- It rejected at least two attempts to amend the Act and immunize state and local governments from awards.22

The Attorney General does not quarrel with the rule established in Fitzpatrick v. Bitzer, supra. Rather, he argues that these plain indications of legislative intent are not enough. In his view, Congress must enact express statutory language making the States liable if it wishes to abrogate their immunity.23 The Attorney General points out that this Court has *695sometimes refused to impose retroactive liability on the States in the absence of an extraordinarily explicit statutory mandate. See Employees v. Missouri Public Health & Welfare Dept., 411 U. S. 279; see also Edelman v. Jordan, 415 U. S. 651. But these eases concern retroactive liability for preliti-gation conduct rather than expenses incurred in litigation seeking only prospective relief.

The Act imposes attorney’s fees “as part of the costs.” Costs have traditionally been awarded without regard for the States’ Eleventh Amendment immunity. The practice of awarding costs against the States goes back to 1849 in this Court. See Missouri v. Iowa, 7 How. 660, 681; North Dakota v. Minnesota, 263 U. S. 583 (collecting cases). The Court has never viewed the Eleventh Amendment as barring such awards, even in suits between States and individual litigants.24

*696In Fairmont Creamery Co. v. Minnesota, 275 U. S. 70, the State challenged this Court’s award of costs, but we squarely-rejected the State’s claim of immunity. Far from requiring an explicit abrogation of state immunity, we relied on a statutory mandate that was entirely silent on the question of state liability.25 The power to make the award was supported by “the inherent authority of the Court in the orderly administration of justice as between all parties litigant.” Id., at 74. A federal court’s interest in orderly, expeditious proceedings “justifies [it] in treating the state just as any other litigant and in imposing costs upon it” when an award is called for. Id., at 77.26

Just as a federal court may treat a State like any other litigant when it assesses costs, so also may Congress amend its definition of taxable costs and have the amended class of costs apply to the States, as it does to all other litigants, without expressly stating that it intends to abrogate the States’ Eleventh Amendment immunity. For it would be absurd to require an express *697reference to state litigants whenever a filing fee, or a new item, such as an expert witness’ fee, is added to the category of taxable costs.27

There is ample precedent for Congress’ decision to authorize an award of attorney’s fees as an item of costs. In England, costs “as between solicitor and client,” Sprague v. Ticonic Nat. Bank, 307 U. S. 161, 167, are routinely taxed today, and have been awarded since 1278. Alyeska Pipeline Service Co. v. Wilderness Society, 421 U. S. 240, 247 n. 18. In America, although fees are not routinely awarded, there are a large number of statutory and common-law situations in which allowable costs include counsel fees.28 Indeed, the federal statutory definition of costs, which was enacted before the Civil War and which remains in effect today, includes certain fixed attorney’s fees as recoverable costs.29 In Fairmont Creamery itself, the Court awarded these statutory attorney’s fees against the *698State of Minnesota along with other taxable costs,30 even though the governing statute said nothing about state liability. It is much too late to single out attorney’s fees as the one kind of litigation cost whose recovery may not be authorized by Congress without an express statutory waiver of the States’ immunity.31

*699Finally, the Attorney General argues that, even if attorney’s fees may be awarded against a State, they should not be awarded in this case, because neither the State nor the Department is expressly named as a defendant. Although the Eleventh Amendment prevented respondents from suing the State by name, their injunctive suit against prison officials was, for all practical purposes, brought against the State. The actions of the Attorney General himself show that. His office has defended this action since it began. See Holt I, 300 F. Supp., at 826. The State apparently paid earlier fee awards; and it was the State’s lawyers who decided to bring this appeal, thereby risking another award.32

*700Like the Attorney General, Congress recognized that suits brought against individual officers for injunctive relief are for all practical purposes suits against the State itself. The legislative history makes it clear that in such suits attorney’s fee awards should generally be obtained “either directly from the official, in his official capacity, from funds of his agency or under his control, or from the State or local government (whether or not the agency or government is a named party) S. Rep. No. 94-1011, p. 5 (1976). Awards against the official in his individual capacity, in contrast, were not to be affected by the statute; in injunctive suits they would continue to be awarded only “under the traditional bad faith standard recognized by the Supreme Court in Alyeska,.” Id., at 5 n. 7. There is no indication in this case that the named defendants litigated in bad faith before the Court of Appeals. Consequently, the Department of Correction is the entity intended by Congress to bear the burden of the counsel-fees award.

The judgment of the Court of Appeals is accordingly affirmed.

It is so ordered.

Mr. Justice Brennan,

concurring.

I join fully in the opinion of the Court and write separately only to answer points made by Mr. Justice Powell.

I agree with the Court that there is no reason in this case to decide more than whether 42 U. S. C. § 1988 (1976 ed.), itself authorizes awards of attorney’s fees against the States. Mr. Justice Powell takes the view, however, that unless 42 U. S. C. § 1983 also authorizes damages awards against the States, the requirements of the Eleventh Amendment are not met. Citing Edelman v. Jordan, 415 U. S. 651 (1974), he concludes that § 1983 does not authorize damages awards against the State and, accordingly, that § 1988 does not either. There are a number of difficulties with this syllogism, but the most striking is its reliance on Edelman v. Jordan, a case whose foundations would seem to have been seriously under*701mined by our later holdings in Fitzpatrick v. Bitzer, 427 U. S. 445 (1976), and Monell v. New York City Dept. of Social Services, 436 U. S. 658 (1978).

It cannot be gainsaid that this Court in Edelman rejected the argument that 42 U. S. C. § 1983 “was intended to create a waiver of a State’s Eleventh Amendment immunity merely because an action could be brought under that section against state officers, rather than against the State itself.” 415 U. S., at 676-677. When Edelman was decided, we had affirmed monetary awards against the States only when they had consented to suit or had waived their Eleventh Amendment immunity. See, e. g., Petty v. Tennessee-Missouri Bridge Comm’n, 359 U. S. 275 (1959); Parden v. Terminal B. Co., 377 U. S. 184 (1964); Employees v. Missouri Public Health & Welfare Dept., 411 U. S. 279 (1973). In Edelman, we summarized the rule of our cases as follows: The “question of waiver or consent under the Eleventh Amendment was found in [our] cases to turn on whether Congress had intended to abrogate the immunity in question, and whether the State by its participation in [a regulated activity] authorized by Congress had in effect consented to the abrogation of [Eleventh Amendment] immunity.” 415 U. S., at 672. At the very least, such consent could not be found unless Congress had authorized suits against “a class of defendants which literally includes States.” Ibid. It was a short jump from that proposition, to the conclusion that § 1983 — which was then thought to include only natural persons among those who could be party defendants, see Monroe v. Pape, 365 U. S. 167, 187-191 (1961) — was not in the class of statutes that might lead to a waiver of Eleventh Amendment immunity. This is best summed up by Me. Justice Rehnquist, the author of Edelman, in his opinion for the Court in Fitzpatrick v. Bitzer, supra:

“We concluded that none of the statutes relied upon by plaintiffs in Edelman contained any authorization by *702Congress to join a State as defendant. The Civil Rights Act of 1871, 42 U. S. C. § 1983, had been held in Monroe v. Pape, 365 U. S. 167, 187-191 (1961), to exclude cities and other municipal corporations from its ambit; that being the case, it could not have been intended to include States as parties defendant.” 427 U. S., at 452.

But time has not stood still. Two Terms ago, we decided Fitzpatrick v. Bitzer, which for the first time in the recent history of the Court asked us to decide “the question of the relationship between the Eleventh Amendment and the enforcement power granted to Congress under § 5 of the Fourteenth Amendment.”1 Id,., at 456. There we concluded that “the Eleventh Amendment, and the principle of state sovereignty which it embodies, . . . are necessarily limited by the enforcement provisions of § 5 of the Fourteenth Amendment.” Ibid. (Citation omitted.) And we went on to hold:

“Congress may, in determining what is 'appropriate legislation’ for the purpose of enforcing the provisions of the Fourteenth Amendment, provide for private suits against States or state officials which are constitutionally impermissible in other contexts.” Ibid.

Then, in Monell v. New York City Dept. of Social Services, supra, decided only weeks ago, we held that the Congress which passed the Civil Rights Act of 1871, now § 1983 — a statute enacted pursuant to § 5 of the Fourteenth Amendment, see 436 IT. S., at 665 — “did intend municipalities and other local government units to be included among those persons to whom § 1983 applies.” Id., at 690. This holding alone would appear to be enough to vitiate the vitality of Fitzpatrick’s explanation of Edelman.2

*703Moreover, central to the holding in Monell was the conclusion that the Act of Feb. 25, 1871, ch. 71, § 2, 16 Stat. 431, provided a definition of the word “person” used to describe the class of defendants in § 1983 suits. 436 U. S., at 688. Although we did not in Monell have to consider whether § 1983 as properly construed makes States liable in damages for their constitutional violations, the conclusion seems inescapable that, at the very least, § 1983 includes among possible defendants “a class . . . which literally includes States.” Edelman v. Jordan, 415 U. S., at 672. This follows immediately from the language of the Act of Feb. 25, 1871:

“[I]n all acts hereafter passed . . . the word 'person’ may extend and be applied to bodies politic and corporate . . . unless the context shows that such words were intended to be used in a more limited sense . . . .”

The phrase “bodies politic and corporate” is now, and certainly would have been in 1871, a synonym for the word “State.” See, e. g., United States v. Maurice, 26 F. Cas. 1211, 1216 (No. 15,747) (CC Va. 1823) (Marshall, C. J.) (“The United States is a government and, consequently, a body politic and corporate”). See also Pfizer Inc. v. Government of India, 434 U. S. 308 (1978).

Given our holding in Monell, the essential premise of our Edelman holding — that no statute involved in Edelman authorized suit against “a class of defendants which literally includes States,” 415 U. S., at 672 — would clearly appear to be no longer true. Moreover, given Fitzpatrick’s holding that Congress has plenary power to make States liable in damages when it acts pursuant to § 5 of the Fourteenth Amendment, it is surely at least an open question whether § 1983 properly construed does not make the States liable for relief of all kinds, notwithstanding the Eleventh Amendment. Whether this is *704in fact so, must of course await consideration in an appropriate case.3

Mr. Justice Powell,

concurring in part and dissenting in part.*

While I join Parts I1 and II-A of the Cburt’s opinion, I cannot subscribe to Part II-B’s reading of the Eleventh Amendment as permitting counsel-fee awards against the State on the authority of a statute that concededly does not effect “an express statutory waiver of the States’ immunity.” Ante, at 698.

Edelman v. Jordan, 415 U. S. 651, 676-677 (1974), rejected the argument that 42 U. S. C. § 1983 “was intended to create a waiver of the State’s Eleventh Amendment immunity merely because an action could be brought under that section against state officers, rather than against the State itself.” In a § 1983 *705action “a federal court’s remedial power, consistent with the Eleventh Amendment, is necessarily limited to prospective injunctive relief, . . . and may not include a retroactive award which requires the payment of funds from the state treasury.” 415 U. S., at 677 (citations omitted). There is no indication in the language of the Civil Rights Attorney’s Fees Awards Act of 1976 (Act), Pub. L. No. 94-559, 90 Stat. 2641, 42 U. S. C. § 1988 (1976 ed.), that Congress sought to overrule that holding.2 In this case, as in Edelman, “the threshold fact of congressional authorization to sue a class of defendants which literally includes States is wholly absent.” 415 U. S., at 672 (emphasis supplied). Absent such authorization, grounded in statutory language sufficiently clear to alert every voting Member of Congress of the constitutional implications of particular legislation, we undermine the values of federalism served by the Eleventh Amendment by inferring from congressional silence an intent to “place new or even enormous fiscal burdens on the States.” Employees v. Missouri Public Health & Welfare Dept., 411 U. S. 279, 284 (1973).

The Court notes that the Committee Reports and the defeat of two proposed amendments indicate a purpose to authorize counsel-fee awards against the States. Ante, at 694. That evidence might provide persuasive support for a finding of “waiver” if this case involved “a congressional enactment which by its terms authorized suit by designated plaintiffs against a general class of defendants which literally included *706States or state instrumentalities.” Edelman, supra, at 672. Compare Fitzpatrick v. Bitzer, 427 U. S. 445, 452 (1976), with Employees, supra, at 283, 284-2S5.3 But in this sensitive area of conflicting interests of constitutional dimension, we should not permit items of legislative history to substitute for explicit statutory language. The Court should be “hesitant to presume .general congressional awareness,” SEC v. Sloan, 436 U. S. 103, 121 (1978), of Eleventh Amendment consequences of a statute that does not make express provision for monetary recovery against the States.4

*707The Court maintains that the Act presents a special case because (i) it imposes attorney’s fees as an element of costs that traditionally have been awarded without regard to the States’ constitutional immunity from monetary liability, and (ii) Congress acted pursuant to its enforcement power under § 5 of the Fourteenth Amendment, as contrasted with its power under more general grants such as the Commerce Clause. I find neither ground a persuasive justification for dilution of the “clear statement” rule.

Notwithstanding the limitations of the Court’s first ground of justification, see ante, at 697 n. 27,1 am unwilling to ignore otherwise applicable principles simply because the statute in question imposes substantial monetary liability as an element of “costs.” Counsel fees traditionally have not been part of the routine litigation expenses assessed against parties in American courts. Cf. Alyeska Pipeline Service Co. v. Wilderness Society, 421 U. S. 240 (1975); Arcambel v. Wiseman, 3 Dall. 306 (1796). Quite unlike those routine expenses, an award of counsel fees may involve substantial sums and is not a charge intimately related to the mechanics of the litigation. I therefore cannot accept the Court’s assumption that counsel-fee awards are part of “the ordinary discipline of the courtroom.” Ante, at 696 n. 24.5

*708Moreover, counsel-fee awards cannot be viewed as having the kind of “ancillary effect on the state treasury,” Edelman, 415 U. S., at 668, that avoids the need for an explicit waiver of Eleventh Amendment protections. As with damages and restitutory relief, an award of counsel fees could impose a substantial burden on the State to make unbudgeted disbursements to satisfy an obligation stemming from past (as opposed to post-litigation) activities. It stretches the rationale of Edelman beyond recognition to characterize such awards as “the necessary result of compliance with decrees which by their terms [are] prospective in nature.” Ibid. In the case of a purely prospective decree, budgeting can take account of the expenditures entailed in compliance, and the State retains some flexibility in implementing the decree, which may reduce the impact on the state fisc. In some situations fiscal considerations may induce the State to curtail the activity triggering the constitutional obligation. Here, in contrast, the State must satisfy a potentially substantial liability without the measure of flexibility that would be available with respect to prospective relief.

The Court’s second ground for application of a diluted “clear statement” rule stems from language in Fitzpatrick recognizing that “[w]hen Congress acts pursuant to § 5” of the Fourteenth Amendment, “it is exercising [legislative] authority under one section of a constitutional Amendment whose other sections by their own terms embody limitations on state authority,” 427 U. S., at 456. I do not view this language as overruling, by implication, Edelmaris holding that no waiver is present in § 1983 6 — the quintessential Fourteenth Amend*709ment measure — or disturbing the vitality of the “threshold [requirement] of congressional authorization to sue a class of defendants which literally includes States,” 415 U. S., at 672.7

*710Because explicit authorization “to join a State as defendant/’ Fitzpatrick, 427 U. S., at 452, is absent here, and because every part of the Act can be given meaning without ascribing to Congress an intention to override the Eleventh Amendment immunity,8 I dissent from Part II-B of the Court’s opinion.

Mr. Justice Rbhnquist,

dissenting.*

The Court’s affirmance of a District Court’s injunction against a prison practice which has not been shown to violate the Constitution can only be considered an aberration in light of decisions as recently as last Term carefully defining the remedial discretion of the federal courts. Dayton Board of Education v. Brinkman, 433 U. S. 406 (1977); Milliken v. Bradley, 433 U. S. 267 (1977) (Milliken II). Nor are any of the several theories which the Court advances in support of its affirmance of the assessment of attorney’s fees against the taxpayers of Arkansas sufficiently convincing to overcome the prohibition of the Eleventh Amendment. Accordingly, I dissent.

*711I

No person of ordinary feeling could fail to be moved by the Court’s recitation of the conditions formerly prevailing in the Arkansas prison system. Yet I fear that the Court has allowed itself to be moved beyond the well-established bounds limiting the exercise of remedial authority by the federal district courts. The purpose and extent of that discretion in another context were carefully defined by the Court’s opinion last Term in Milliken II, swpra, at 280-281:

“In the first place, like other equitable remedies, the nature of the desegregation remedy is to be determined by the nature and scope of the constitutional violation. Swann v. Charlotte-Mecklenburg Board of Education, 402 U. S. [1,] 16 [(1971)]. The remedy must therefore be related to The condition alleged to offend the Constitution . . . .’ Milliken [v. Bradley], 418 U. S. [717,] 738 [(1974)]. Second, the decree must indeed be remedial in nature, that is, it must be designed as nearly as possible To restore the victims of discriminatory conduct to the position they would have occupied in the absence of such conduct.’ Id., at 746. Third, the federal courts in devising a remedy must take into account the interests of state and local authorities in managing their own affairs, consistent with the Constitution.” (Footnotes omitted.) 1

*712The District Court’s order limiting the maximum period of punitive isolation to 30 days in no way relates to any condition found offensive to the Constitution. It is, when stripped of descriptive verbiage, a prophylactic rule, doubtless well designed to assure a more humane prison system in Arkansas, but not complying with the limitations set forth in Milliken II, supra. Petitioners do not dispute the District Court’s conclusion that the overcrowded conditions and the inadequate diet provided for those prisoners in punitive isolation offended the Constitution, but the District Court has ordered a cessation of those practices. The District Court found that the confinement of two prisoners in a single cell on a restricted diet for 30 days did not violate the Eighth Amendment. 410 F. Supp. 251, 278 (ED Ark. 1976). While the Court today remarks that “the length of confinement cannot be ignored,” ante, at 686, it does not find that confinement under the conditions described by the District Court becomes unconstitutional on the 31st day. It must seek other justifications for its affirmance of that portion of the District Court’s order.

Certainly the provision is not remedial in the sense that it “restore[s] the victims of discriminatory conduct to the position they would have occupied in the absence of such conduct.” Milliken v. Bradley, 418 U. S. 717, 746 (1974) (.Milliken I). The sole effect of the provision is to grant future offenders against prison discipline greater benefits than the Constitution requires; it does nothing to remedy the plight of past victims of conditions which may well have been unconstitutional. A prison is unlike a school system, in which students in the later grades may receive special instruction to compensate for discrimination to which they were subjected in the *713earlier grades. Milliken II, supra, at 281-283. Nor has it been shown that petitioners’ conduct had any collateral effect upon private actions for which the District Court may seek to compensate so as to eliminate the continuing effect of past unconstitutional conduct. See Swann v. Charlotte-Mecklenburg Board of Education, 402 U. S. 1, 28 (1971). Even where such remedial relief is justified, a district court may go no further than is necessary to eliminate the consequences of official unconstitutional conduct. Dayton, supra, at 419-420; Pasadena Board of Education v. Spangler, 427 U. S. 424, 435-437 (1976) ; Swann, supra, at 31-32.

The Court’s only asserted justification for its affirmance of the decree, despite its dissimilarity to remedial decrees in other contexts, is that it is “a mechanical — and therefore an easily enforced — method of minimizing overcrowding.” Ante, at 688 n. 11. This conclusion fails adequately to take into account the third consideration cited in Milliken II: “the interests of state and local authorities in managing their own affairs, consistent with the Constitution.” 433 U. S., at 281. The prohibition against extended punitive isolation, a practice which has not been shown to be inconsistent with the Constitution, can only be defended because of the difficulty of policing the District Court’s explicit injunction against the overcrowding and inadequate diet which have been found to be violative of the Constitution. But even if such an expansion of remedial authority could be justified in a case where the defendants had been repeatedly contumacious, this is not such a case. The District Court’s dissatisfaction with petitioners’ performance under its earlier direction to “make a substantial start,” Holt v. Sarver, 300 F. Supp. 825, 833 (ED Ark. 1969), on alleviating unconstitutional conditions cannot support an inference that petitioners are prepared to defy the specific orders now laid down by the District Court and not challenged by the petitioners. A proper respect for “the interests of state and local authorities in managing their own *714affairs,” Milliken II, 433 U. S., at 281, requires the opposite conclusion.2

The District Court's order enjoins a practice which has not been found inconsistent with the Constitution. The only ground for the injunction, therefore, is the prophylactic one of assuring that no unconstitutional conduct will occur in the future. In a unitary system of prison management there would be much to be said for such a rule, but neither this Court nor any other federal court is entrusted with such a management role under the Constitution.

II

The Court advances separate theories to support the separate awards of attorney’s fees in this case. First, the Court holds that the taxpayers of Arkansas may be held responsible for the bad faith of their officials in the litigation before the District Court. Second, it concludes that the award of fees in the Court of Appeals, where there was no bad faith, is authorized by the Civil Rights Attorney’s Fees Awards Act of 1976. Pub. L. No. 94-559, 90 Stat. 2641, 42 U. S. C. § 1988 (1976 ed.). The first holding results in a totally unnecessary intrusion upon the State’s conduct of its own affairs, and the second is not supportable under this Court’s earlier decisions outlining congressional authority to abrogate the protections of the Eleventh Amendment.

A

Petitioners do not contest the District Court’s finding that they acted in bad faith. For this reason, the Court has no *715occasion to address the nature of the showing necessary to support an award of attorney’s fees for bad faith under Alyeska Pipeline Service Co. v. Wilderness Society, 421 U. S. 240, 258-259 (1975). The only issue before us is whether a proper finding of bad faith on the part of state officials will support an award of attorney’s fees directly against the state treasury under the ancillary-effect doctrine of Edelman v. Jordan, 415 U. S. 651, 668 (1974).

The ancillary-effect doctrine recognized in Edelman is a necessary concomitant of a federal court’s authority to require state officials to conform their conduct to the dictates of the Constitution. “State officials, in order to shape their official conduct to the mandate of the Court’s decrees, would more likely have to spend money from the state treasury than if they had been left free to pursue their previous course of conduct.” Id., at 668. The Court today suggests that a federal court may impose a retroactive financial penalty upon a State when it fails to comply with prospective relief previously and validly ordered. “If a state agency refuses to adhere to a court order, a financial penalty may be the most effective means of insuring compliance.” Ante, at 691. This application oí the ancillary-effect doctrine has never before been recognized by this Court, and there is no need to do so in this case, since it has not been shown that these petitioners have “refuse[d] to adhere to a court order.” A State’s jealous defense of its authority to operate its own correctional system cannot casually be equated with contempt of court.3

*716Even were I to agree with the Court that petitioners had willfully defied federal decrees, I could not conclude that the award of fees against the taxpayers of Arkansas would be justified, since there is a less intrusive means of insuring respondents’ right to relief. It is sufficient to order an award of fees against those defendants, acting in their official capacity, who are personally responsible for the recalcitrance which the District Court wishes to penalize. There is no reason for the federal courts to engage in speculation as to whether the imposition of a fine against the State is “less intrusive” than “sending high state officials to jail.” Ibid. So long as the rights of the plaintiffs and the authority of the District Court are amply vindicated by an award of fees, it should be a matter of no concern to the court whether those fees are paid by state officials personally or by the State itself. The Arkansas Legislature has already made statutory provision for deciding when its officials shall be reimbursed by the State for judgments ordered by the federal courts. 1977 Ark. Gen. Act No. 543.

The Court presents no persuasive reason for its conclusion that the decision of who must pay such fees may not safely be left to the State involved. It insists, ante, at 699 n. 32, that it is “manifestly unfair” to leave the individual state officers to pay the award of counsel fees rather than permitting their collection directly from the state treasury. But petitioners do not contest the District Court’s finding that they acted in bad faith, and thus the Court’s insistence that it is “unfair” to impose attorney’s fees on them individually rings somewhat hollow.4 Even in a case where the equities were more strongly in favor of the individual state officials (as opposed to the State as an entity) than they are in this case, *717the possibility of individual liability in damages of a state official where the State itself could not be held liable is as old as Ex parte Young, 209 U. S. 123 (1908), and has been repeatedly reaffirmed by decisions of this Court. Great Northern Lije Insurance Co. v. Read, 322 U. S. 47 (1944); Ford Motor Co. v. Department of Treasury, 323 U. S. 459 (1945); Edelman v. Jordan, supra. Since the Court evidences no disagreement with this line of cases, its assertion of “unfairness” is not only doubtful in fact but also irrelevant as a matter of law. Likewise, the Court’s fear that imposition of liability would inhibit state officials in the fearless exercise of their duties may be remedied, if deemed desirable, by legislation in each of the various States similar to that which Arkansas has already enacted.

B

For the reasons stated in the dissenting portion of my Brother Powell’s opinion, which I join, I do not agree that the Civil Rights Attorney’s Fees Awards Act of 1976 can be considered a valid congressional abrogation of the State’s Eleventh Amendment immunity. I have in addition serious reservations about the lack of any analysis accompanying the Court’s transposition of the holding of Fitzpatrick v. Bitzer, 427 U. S. 445 (1976), to this case. In Fitzpatrick, we held that under § 5 of the Fourteenth Amendment Congress could explicitly allow for recovery against state agencies without violating the Eleventh Amendment. But in Fitzpatrick, supra, there was conceded to be a violation of the Equal Protection Clause which is contained in haec verba in the language of the Fourteenth Amendment itself. In this case the claimed constitutional violation is the infliction of cruel and unusual punishment, which is expressly prohibited by the Eighth but not by the Fourteenth Amendment. While the Court has held that the Fourteenth Amendment “incorporates” the prohibition against cruel and unusual punishment, it is not at all clear to me that it follows that Congress has the same enforcement power *718under § 5 with respect to a constitutional provision which has merely been judicially “incorporated” into the Fourteenth Amendment that it has with respect to a provision which was placed in that Amendment by the drafters.

I would therefore reverse the judgment of the Court of Appeals in its entirety.

*60Justice White

delivered the opinion of the Court.

This case presents the question whether a State, or an official of the State while acting in his or her official capacity, is a “person” within the meaning of Rev. Stat. § 1979, 42 U. S. C. § 1983.

Petitioner Ray Will filed suit in Michigan Circuit Court alleging various violations of the United States and Michigan Constitutions as grounds for a claim under §1983.1 He alleged that he had been denied a promotion to a data systems analyst position with the Department of State Police for an improper reason, that is, because his brother had been a student activist and the subject of a “red squad” file maintained by respondent. Named as defendants were the Department of State Police and the Director of State Police in his official capacity, also a respondent here.2

The Circuit Court remanded the case to the Michigan Civil Service Commission for a grievance hearing. While the grievance was pending, petitioner filed suit in the Michigan *61Court of Claims raising an essentially identical § 1983 claim. The Civil Service Commission ultimately found in petitioner’s favor, ruling that respondents had refused to promote petitioner because of “partisan considerations.” App. 46. On the basis of that finding, the state-court judge, acting in both the Circuit Court and the Court of Claims cases, concluded that petitioner had established a violation of the United States Constitution. The judge held that the Circuit Court action was barred under state law but that the Claims Court action could go forward. The judge also ruled that respondents were persons for purposes of § 1983.

The Michigan Court of Appeals vacated the judgment against the Department of State Police, holding that a State is not a person under § 1983, but remanded the case for determination of the possible immunity of the Director of State Police from liability for damages. The Michigan Supreme Court granted discretionary review and affirmed the Court of Appeals in part and reversed in part. Smith v. Department of Pub. Health, 428 Mich. 540, 410 N. W. 2d 749 (1987). The Supreme Court agreed that the State itself is not a person under § 1983, but held that a state official acting in his or her official capacity also is not such a person.

The Michigan Supreme Court’s holding that a State is not a person under § 1983 conflicts with a number of state- and federal-court decisions to the contrary.3 We granted certio-rari to resolve the conflict. 485 U. S. 1005 (1988).

*62Prior to Monell v. New York City Dept. of Social Services, 436 U. S. 668 (1978), the question whether a State is a person within the meaning of § 1983 had been answered by this Court in the negative. In Monroe v. Pape, 365 U. S. 167, 187-191 (1961), the Court had held that a municipality was not a person under § 1983. “[T]hat being the case,” we reasoned, § 1983 “could not have been intended to include States as parties defendant.” Fitzpatrick v. Bitzer, 427 U. S. 446, 452 (1976).

But in Monell, the Court overruled Monroe, holding that a municipality was a person under § 1983. 436 U. S., at 690. Since then, various members of the Court have debated whether a State is a person within the meaning of § 1983, see Hutto v. Finney, 437 U. S. 678, 700-704 (1978) (Brennan, J., concurring); id., at 708, n. 6 (Powell, J., concurring in *63part and dissenting in part), but this Court has never expressly dealt with that issue.4

Some courts, including the Michigan Supreme Court here, have construed our decision in Quern v. Jordan, 440 U. S. 332 (1979), as holding by implication that a State is not a person under § 1983. See Smith v. Department of Pub. Health, supra, at 581, 410 N. W. 2d, at 767. See also, e. g., State v. Green, 633 P. 2d 1381, 1382 (Alaska 1981); Woodbridge v. Worcester State Hospital, 384 Mass. 38, 44-45, n. 7, 423 N. E. 2d 782, 786, n. 7 (1981); Edgar v. State, 92 Wash. 2d 217, 221, 595 P. 2d 534, 537 (1979), cert. denied, 444 U. S. 1077 (1980). Quern held that §1983 does not override a State’s Eleventh Amendment immunity, a holding that the concurrence suggested was “patently dicta” to the effect that a State is not a person, 440 U. S., at 350 (Brennan, J., concurring in judgment).

Petitioner filed the present § 1983 actions in Michigan state court, which places the question whether a State is a person under § 1983 squarely before us since the Eleventh Amend*64ment does not apply in state courts. Maine v. Thiboutot, 448 U. S. 1, 9, n. 7 (1980). For the reasons that follow, we reaffirm today what we had concluded prior to Monell and what some have considered implicit in Quern: that a State is not a person within the meaning of § 1983.

We observe initially that if a State is a “person” within the meaning of § 1983, the section is to be read as saying that “every person, including a State, who, under color of any statute, ordinance, regulation, custom, or usage, of any State or Territory or the District of Columbia, subjects . . . .” That would be a decidedly awkward way of expressing an intent to subject the States to liability. At the very least, reading the statute in this way is not so clearly indicated that it provides reason to depart from the often-expressed understanding that “fin common usage, the term ‘person’ does not include the sovereign, [and] statutes employing the [word] are ordinarily construed to exclude it.’” Wilson v. Omaha Tribe, 442 U. S. 653, 667 (1979) (quoting United States v. Cooper Corp., 312 U. S. 600, 604 (1941)). See also United States v. Mine Workers, 330 U. S. 258, 275 (1947).

This approach is particularly applicable where it is claimed that Congress has subjected the States to liability to which they had not been subject before. In Wilson v. Omaha Tribe, supra, we followed this rule in construing the phrase “white person” contained in 25 U. S. C. § 194, enacted as Act of June 30, 1834, 4 Stat. 729, as not including the “sovereign States of the Union.” 442 U. S., at 667. This common usage of the term “person” provides a strong indication that “person” as used in § 1983 likewise does not include a State.5

*65The language of § 1983 also falls far short of satisfying the ordinary rule of statutory construction that if Congress intends to alter the “usual constitutional balance between the States and the Federal Government,” it must make its intention to do so “unmistakably clear in the language of the statute.” Atascadero State Hospital v. Scanlon, 473 U. S. 234, 242 (1985); see also Pennhurst State School and Hospital v. Halderman, 465 U. S. 89, 99 (1984). Atascadero was an Eleventh Amendment case, but a similar approach is applied in other contexts. Congress should make its intention “clear and manifest” if it intends to pre-empt the historic powers of the States, Rice v. Santa Fe Elevator Corp., 331 U. S. 218, 230 (1947), or if it intends to impose a condition on the grant of federal moneys, Pennhurst State School and Hospital v. Halderman, 451 U. S. 1, 16 (1981); South Dakota v. Dole, 483 U. S. 203, 207 (1987). “In traditionally sensitive areas, such as legislation affecting the federal balance, the requirement of clear statement assures that the legislature has in fact faced, and intended to bring into issue, the critical matters involved in the judicial decision.” United States v. Bass, 404 U. S. 336, 349 (1971).

Our conclusion that a State is not a “person” within the meaning of § 1983 is reinforced by Congress’ purpose in en*66acting the statute. Congress enacted § 1 of the Civil Rights Act of 1871, 17 Stat. 13, the precursor .to § 1983, shortly after the end of the Civil War “in response to the widespread deprivations of civil rights in the Southern States and the inability or unwillingness of authorities in those States to protect those rights or punish wrongdoers.” Felder v. Casey, 487 U. S. 131, 147 (1988). Although Congress did not establish federal courts as the exclusive forum to remedy these deprivations, ibid., it is plain that “Congress assigned to the federal courts a paramount role” in this endeavor, Patsy v. Board of Regents of Florida, 457 U. S. 496, 503 (1982).

Section 1983 provides a federal forum to remedy many deprivations of civil liberties, but it does not provide a federal forum for litigants who seek a remedy against a State for alleged deprivations of civil liberties. The Eleventh Amendment bars such suits unless the State has waived its immunity, Welch v. Texas Dept. of Highways and Public Transportation, 483 U. S. 468, 472-473 (1987) (plurality opinion), or unless Congress has exercised its undoubted power under § 5 of the Fourteenth Amendment to override that immunity. That Congress, in passing § 1983, had no intention to disturb the States’ Eleventh Amendment immunity and so to alter the federal-state balance in that respect was made clear in our decision in Quern. Given that a principal purpose behind the enactment of § 1983 was to provide a federal forum for civil rights claims, and that Congress did not provide such a federal forum for civil rights claims against States, we cannot accept petitioner’s argument that Congress intended nevertheless to create a cause of action against States to be brought in state courts, which are precisely the courts Congress sought to allow civil rights claimants to avoid through § 1983.

This does not mean, as petitioner suggests, that we think that the scope of the Eleventh Amendment and the scope of § 1983 are not separate issues. Certainly they are. But in deciphering congressional intent as to the scope of § 1983, the *67scope of the Eleventh Amendment is a consideration, and we decline to adopt a reading of § 1983 that disregards it.6

Our conclusion is further supported by our holdings that in enacting §1983, Congress did not intend to override well-established immunities or defenses under the common law. “One important assumption underlying the Court’s decisions in this area is that members of the 42d Congress were familiar with common-law principles, including defenses previously recognized in ordinary tort litigation, and that they likely intended these common-law principles to obtain, absent specific provisions to the contrary.” Newport v. Fact Concerts, Inc., 453 U. S. 247, 258 (1981). Stump v. Sparkman, 435 U. S. 349, 356 (1978); Scheuer v. Rhodes, 416 U. S. 232, 247 (1974); Pierson v. Ray, 386 U. S. 547, 554 (1967); and Tenney v. Brandhove, 341 U. S. 367, 376 (1951), are also to this effect. The doctrine of sovereign immunity was a familiar doctrine at common law. “The principle is elementary that a State cannot be sued in its own courts without its consent.” Railroad Co. v. Tennessee, 101 U. S. 337, 339 (1880). It is an “established principle of jurisprudence” that the sovereign cannot be sued in its own courts without its consent. Beers v. Arkansas, 20 How. 527, 529 (1858). We cannot conclude that § 1983 was intended to disregard the well-established immunity of a State from being sued without its consent.7

*68The legislative history of § 1983 does not suggest a different conclusion. Petitioner contends that the congressional debates on § 1 of the 1871 Act indicate that § 1983 was intended to extend to the full reach of the Fourteenth Amendment and thereby to provide a remedy “ ‘against all forms of official violation of federally protected rights.”’ Brief for Petitioner 16 (quoting Monell, 436 U. S., at 700-701). He refers us to various parts of the vigorous debates accompanying the passage of § 1983 and revealing that it was the failure of the States to take appropriate action that was undoubtedly the motivating force behind § 1983. The inference must be drawn, it is urged, that Congress must have intended to subject the States themselves to liability. But the intent of Congress to provide a remedy for unconstitutional state action does not without more include the sovereign States among those persons against whom § 1983 actions would lie. Construing § 1983 as a remedy for “official violation of federally protected rights” does no more than confirm that the section is directed against state action — action “under color of” state law. It does not suggest that the State itself was a person that Congress intended to be subject to liability.

Although there were sharp and heated debates, the discussion of § 1 of the bill, which contained the present § 1983, was not extended. And although in other respects the impact on state sovereignty was much talked about, no one suggested that § 1 would subject the States themselves to a damages suit under federal law. Quern, 440 U. S., at 343. There was complaint that § 1 would subject state officers to damages liability, but no suggestion that it would also expose the States themselves. Cong. Globe, 42d Cong., 1st Sess., *69366, 385 (1871). We find nothing substantial in the legislative history that leads us to believe that Congress intended that the word “person” in § 1983 included the States of the Union. And surely nothing in the debates rises to the clearly expressed legislative intent necessary to permit that construction.

Likewise, the Act of Feb. 25, 1871, §2, 16 Stat. 431 (the “Dictionary Act”),8 on which we relied in Monell, supra, at 688-689, does not counsel a contrary conclusion here. As we noted in Quern, that Act, while adopted prior to § 1 of the Civil Rights Act of 1871, was adopted after §2 of the Civil Rights Act of 1866, from which § 1 of the 1871 Act was derived. 440 U. S., at 341, n. 11. Moreover, we disagree with Justice Brennan that at the time the Dictionary Act was passed “the phrase ‘bodies politic and corporate’ was understood to include the States.” Post, at 78. Rather, an examination of authorities of the era suggests that the phrase was used to mean corporations, both private and public (municipal), and not to include the States.9 In our view, the *70Dictionary Act, like § 1983 itself and its legislative history, fails to evidence a clear congressional intent that States be held liable.

Finally, Monell itself is not to the contrary. True, prior to Monell the Court had reasoned that -if municipalities were not persons then surely States also were not. Fitzpatrick v. Bitzer, 427 U. S., at 452. And Monell overruled Monroe, undercutting that logic. But it does not follow that if municipalities are persons then so are States. States are protected by the Eleventh Amendment while municipalities are not, Monell, 436 U. S., at 690, n. 54, and we consequently limited our holding in Monell “to local government units which are not considered part of the State for Eleventh Amendment purposes,” ibid. Conversely, our holding here does not cast any doubt on Monell, and applies only to States or governmental entities that are considered “arms of the State” for Eleventh Amendment purposes. See, e. g., Mt. Healthy Bd. of Ed. v. Doyle, 429 U. S. 274, 280 (1977).

Petitioner asserts, alternatively, that state officials should be considered “persons” under § 1983 even though acting in their official capacities. In this case, petitioner named as defendant not only the Michigan Department of State Police but also the Director of State Police in his official capacity.

*71Obviously, state officials literally are persons. But a suit against a state official in his or her official capacity is not a suit against the official but rather is a suit against the official’s office. Brandon v. Holt, 469 U. S. 464, 471 (1985). As such, it is no different from a suit against the State itself. See, e. g., Kentucky v. Graham, 473 U. S. 159, 165-166 (1985); Monell, supra, at 690, n. 55. We see no reason to adopt a different rule in the present context, particularly when such a rule would allow petitioner to circumvent congressional intent by a mere pleading device.10

We hold that neither a State nor its officials acting in their official capacities are “persons” under § 1983. The judgment of the Michigan Supreme Court is affirmed.

It is so ordered.

Justice Brennan,

dissenting.

Because this case was brought in state court, the Court concedes, the Eleventh Amendment is inapplicable here. See ante, at 63-64. Like the guest who would not leave, *72however, the Eleventh Amendment lurks everywhere in today’s decision and, in truth, determines its outcome.

1 — (

Section 1 of the Civil Rights Act of 1871, 42 U. S. C. § 1983, renders certain “persons” liable for deprivations of constitutional rights. The question presented is whether the word “person” in this statute includes the States and state officials acting in their official capacities.

One might expect that this statutory question would generate a careful and thorough analysis of the language, legislative history, and general background of § 1983. If this is what one expects, however, one will be disappointed by today’s decision. For this case is not decided on the basis of our ordinary method of statutory construction; instead, the Court disposes of it by means of various rules of statutory interpretation that it summons to its aid each time the question looks close. Specifically, the Court invokes the following interpretative principles: the word “persons” is ordinarily construed to exclude the sovereign; congressional intent to affect the federal-state balance must be “clear and manifest”; and intent to abrogate States’ Eleventh Amendment immunity must appear in the language of the statute itself. The Court apparently believes that each of these rules obviates the need for close analysis of a statute’s language and history. Properly applied, however, only the last of these interpretative principles has this effect, and that principle is not pertinent to the case before us.

The Court invokes, first, the “often-expressed understanding” that “ ‘in common usage, the term “person” does not include the sovereign, [and] statutes employing the [word] are ordinarily construed to exclude it.’” Ante, at 64, quoting Wilson v. Omaha Tribe, 442 U. S. 653, 667 (1979). This rule is used both to refute the argument that the language of §1983 demonstrates an intent that States be included as defendants, ante, at 64, and to overcome the argu*73ment based on the Dictionary Act’s definition of “person” to include bodies politic and corporate, ante, at 69-70. It is ironic, to say the least, that the Court chooses this interpretive rule in explaining why the Dictionary Act is not decisive, since the rule is relevant only when the word “persons” has no statutory definition. When one considers the origins and content of this interpretive guideline, moreover, one realizes that it is inapplicable here and, even if applied, would defeat rather than support the Court’s approach and result.

The idea that the word “persons” ordinarily excludes the sovereign can be traced to the “familiar principle that the King is not bound by any act of Parliament unless he be named therein by special and particular words.” Dollar Savings Bank v. United States, 19 Wall. 227, 239 (1874). As this passage suggests, however, this interpretive principle applies only to “the enacting sovereign.” United States v. California, 297 U. S. 175, 186 (1936). See also Jefferson County Pharmaceutical Assn., Inc. v. Abbott Laboratories, 460 U. S. 150, 161, n. 21 (1983). Furthermore, as explained in United States v. Herron, 20 Wall. 251, 255 (1874), even the principle as applied to the enacting sovereign is not without limitations: “Where an act of Parliament is made for the public good, as for the advancement of religion and justice or to prevent injury and wrong, the king is bound by such act, though not particularly named therein; but where a statute is general, and thereby any prerogative, right, title, or interest is divested or taken from the king, in such case the king is not bound, unless the statute is made to extend to him by express words.” It would be difficult to imagine a statute more clearly designed “for the public good,” and “to prevent injury and wrong,” than § 1983.

Even if this interpretive principle were relevant to this case, the Court’s invocation of it to the exclusion of careful statutory analysis is in error. As we have made clear, this principle is merely “an aid to consistent construction of statutes of the enacting sovereign when their purpose is in *74doubt, but it does not require that the aim of a statute fairly to be inferred be disregarded because not explicitly stated.” United States v. California, supra, at 186. Indeed, immediately following the passage quoted by the Court today, ante, at 64, to the effect that statutes using the word “person” are “ordinarily construed to exclude” the sovereign, we stated:

“But there is no hard and fast rule of exclusion. The purpose, the subject matter, the context, the legislative history, and the executive interpretation of the statute are aids to construction which may indicate an intent, by the use of the term, to bring state or nation within the scope of the law.

“Decision is not to be reached by a strict construction of the words of the Act, nor by the application of artificial canons of construction. On the contrary, we are to read the statutory language in its ordinary and natural sense, and if doubts remain, resolve them in the light, not only of the policy intended to be served by the enactment, but, as well, by all other available aids to construction.” United States v. Cooper Corp., 312 U. S. 600, 604-605 (1941).

See also Wilson v. Omaha Indian Tribe, supra, at 667 (“There is . . . ‘no hard and fast rule of exclusion,’ United States v. Cooper Corp., [312 U. S. 600,] 604-605 [(1941)]; and much depends on the context, the subject matter, legislative history, and executive interpretation”); Pfizer Inc. v. India, 434 U. S. 308, 315-318 (1978); Guarantee Title & Trust Co. v. Title Guaranty & Surety Co., 224 U. S. 152, 155 (1912); Lewis v. United States, 92 U. S. 618, 622 (1875); Green v. United States, 9 Wall. 655, 658 (1870).

The second interpretive principle that the Court invokes comes from cases such as Rice v. Santa Fe Elevator Corp., 331 U. S. 218, 230 (1947); Pennhurst State School and Hospital v. Halderman, 451 U. S. 1, 16 (1981); South Dakota v. Dole, 483 U. S. 203, 207-208 (1987); and United States v. *75Bass, 404 U. S. 336, 349 (1971), which require a “clear and manifest” expression of congressional intent to change some aspect of federal-state relations. Ante, at 65. These cases do not, however, permit substitution of an absolutist rule of statutory construction for thorough statutory analysis. Indeed, in each of these decisions the Court undertook a careful and detailed analysis of the statutory language and history under consideration. Rice is a particularly inapposite source for the interpretive method that the Court today employs, since it observes that, according to conventional pre-emption analysis, a “clear and manifest” intent to pre-empt state legislation may appear in the “scheme” or “purpose” of the federal statute. See 331 U. S., at 230.

The only principle of statutory construction employed by the Court that would justify a perfunctory and inconclusive analysis of a statute’s language and history is one that is irrelevant to this case. This is the notion “that if Congress intends to alter the ‘usual constitutional balance between the States and the Federal Government,’ it must make its intention to do so ‘unmistakably clear in the language of the statute.’” Ante, at 65, quoting Atascadero State Hospital v. Scanlon, 473 U. S. 234, 242 (1985). As the Court notes, Atascadero was an Eleventh Amendment case; the “constitutional balance” to which Atascadero refers is that struck by the Eleventh Amendment as this Court has come to interpret it. Although the Court apparently wishes it were otherwise, the principle of interpretation that Atascadero announced is unique to cases involving the Eleventh Amendment.

Where the Eleventh Amendment applies, the Court has devised a clear-statement principle more robust than its requirement of clarity in any other situation. Indeed, just today, the Court has intimated that this clear-statement principle is not simply a means of discerning congressional intent. See Dellmuth v. Muth, post, at 232 (concluding that one may not rely on a “permissible inference” from a statute’s language and structure in finding abrogation of immunity); post, *76at 238-239 (Brennan, J., dissenting); but see Pennsylvania v. Union Gas Co., ante, p. 1. Since this case was brought in state court, however, this strict drafting requirement has no application here. The Eleventh Amendment can hardly be “a consideration,” ante, at 67, in a suit to which it does not apply.

That this Court has generated a uniquely daunting requirement of clarity in Eleventh Amendment cases explains why Quern v. Jordan, 440 U. S. 332 (1979), did not decide the question before us today. B ecause only the Eleventh Amendment permits use of this clear-statement principle, the holding of Quern v. Jordan that § 1983 does not abrogate States’ Eleventh Amendment immunity tells us nothing about the meaning of the term “person” in § 1983 as a matter of ordinary statutory construction. Quern’s conclusion thus does not compel, or even suggest, a particular result today.

The singularity of this Court’s approach to statutory interpretation in Eleventh Amendment cases also refutes the Court’s argument that, given Quern’s holding, it would make no sense to construe § 1983 to include States as “persons.” See ante, at 66. This is so, the Court suggests, because such a construction would permit suits against States in state but not federal court, even though a major purpose of Congress in enacting § 1983 was to provide a federal forum for litigants who had been deprived of their constitutional rights. See, e. g., Monroe v. Pape, 365 U. S. 167 (1961). In answering the question whether § 1983 provides a federal forum for suits against the States themselves, however, one must apply the clear-statement principle reserved for Eleventh Amendment cases. Since this principle is inapplicable to suits brought in state court, and inapplicable to the question whether States are among those subject to a statute, see Employees v. Missouri Dept. of Public Health and Welfare, 411 U. S. 279, 287 (1973); Atascadero, supra, at 240, n. 2, the answer to the question whether §1983 provides a federal forum for suits against the States may be, and most often will *77be, different from the answer to the kind of question before us today. Since the question whether Congress has provided a federal forum for damages suits against the States is answered by applying a uniquely strict interpretive principle, see supra, at 75, the Court should not pretend that we have, in Quern, answered the question whether Congress intended to provide a federal forum for such suits, and then reason backwards from that “intent” to the conclusion that Congress must not have intended to allow such suits to proceed in state court.

In short, the only principle of statutory interpretation that permits the Court to avoid a careful and thorough analysis of § 1983’s language and history is the clear-statement principle that this Court has come to apply in Eleventh Amendment cases — a principle that is irrelevant to this state-court action. In my view, a careful and detailed analysis of § 1983 leads to the conclusion that States are “persons” within the meaning of that statute.

II

Section 1983 provides:

“Every person who, under color of any statute, ordinance, regulation, custom, or usage, of any State or Territory or the District of Columbia, subjects, or causes to be subjected, any citizen of the United States or other person within the jurisdiction thereof to the deprivation of any rights, privileges, or immunities secured by the Constitution and laws, shall be liable to the party injured in an action at law, suit in equity, or other proper proceeding for redress.”

Although § 1983 itself does not define the term “person,” we are not without a statutory definition of this word. “Any analysis of the meaning of the word ‘person’ in § 1983 . . . must begin . . . with the Dictionary Act.” Monell v. New York City Dept. of Social Services, 436 U. S. 658, 719 (1978) (Rehnquist, J., dissenting). Passed just two months be*78fore § 1983, and designed to “suppl[y] rules of construction for all legislation,” ibid., the Dictionary Act provided:

“That in all acts hereafter passed . . . the word ‘person’ may extend and be applied to bodies politic and corporate . . . unless the context shows that such words were intended to be used in a more limited sense . . . Act of Feb. 25, 1871, §2, 16 Stat. 431.

In Monell, we held this definition to be not merely allowable but mandatory, requiring that the word “person” be construed to include “bodies politic and corporate” unless the statute under consideration “by its terms called for a deviation from this practice.” 436 U. S., at 689-690, n. 53. Thus, we concluded, where nothing in the “context” of a particular statute “call[s] for a restricted interpretation of the word ‘person,’ the language of that [statute] should prima facie be construed to include ‘bodies politic’ among the entities that could be sued.” Ibid.

Both before and after the time when the Dictionary Act and § 1983 were passed, the phrase “bodies politic and corporate” was understood to include the States. See, e. g., J. Bouvier, 1 A Law Dictionary Adapted to the Constitution and Laws of the United States of America 185 (11th ed. 1866); W. Shumaker & G. Longsdorf, Cyclopedic Dictionary of Law 104 (1901); Chisholm v. Georgia, 2 Dali. 419, 447 (1793) (Iredell, J.); id., at 468 (Cushing, J.); Cotton v. United States, 11 How. 229, 231 (1851) (“Every sovereign State is of necessity a body politic, or artificial person”); Poindexter v. Greenhow, 114 U. S. 270, 288 (1885); McPherson v. Blacker, 146 U. S. 1, 24 (1892); Heim v. McCall, 239 U. S. 175, 188 (1915). See also United States v. Maurice, 2 Brock. 96, 109 (CC Va. 1823) (Marshall, C. J.) (“The United States is a government, and, consequently, a body politic and corporate”); Van Brocklin v. Tennessee, 117 U. S. 151, 154 (1886) (same). Indeed, the very legislators who passed § 1 referred to States in these terms. See, e. g., Cong. Globe, 42d Cong., 1st Sess., 661-662 (1871) (Sen. Vickers) (“What is a State? Is *79it not a body politic and corporate?”); id., at 696 (Sen. Ed-munds) (“A State is a corporation”).

The reason why States are “bodies politic and corporate” is simple: just as a corporation is an entity that can act only through its agents, “[t]he State is a political corporate body, can act only through agents, and can command only by laws.” Poindexter v. Greenhow, supra, at 288. See also Black’s Law Dictionary 159 (5th ed. 1979) (“[B]ody politic or corporate”: “A social compact by which the whole people covenants with each citizen, and each citizen with the whole people, that all shall be governed by certain laws for the common good”). As a “body politic and corporate,” a State falls squarely within the Dictionary Act’s definition of a “person.”

While it is certainly true that the phrase “bodies politic and corporate” referred to private and public corporations, see ante, at 69, and n. 9, this fact does not draw into question the conclusion that this phrase also applied to the States. Phrases may, of course, have multiple referents. Indeed, each and every dictionary cited by the Court accords a broader realm — one that comfortably, and in most cases explicitly, includes the sovereign — to this phrase than the Court gives it today. See 1 B. Abbott, Dictionary of Terms and Phrases Used in American or English Jurisprudence 155 (1879) (“[T]he term body politic is often used in a general way, as meaning the state or the sovereign power, or the city government, without implying any distinct express incorporation”); W. Anderson, A Dictionary of Law 127 (1893) (“[B]ody politic”: “The governmental, sovereign power: a city or a State”); Black’s Law Dictionary 143 (1891) (“[Bjody politic”: “It is often used, in a rather loose way, to designate the state or nation or sovereign power, or the government of a county or municipality, without distinctly connoting any express and individual corporate charter”); 1 A. Burrill, A Law Dictionary and Glossary 212 (2d ed. 1871) (“[B]ody politic”: “A body to take in succession, framed by policy”', “[pjarticu-*80larly applied, in the old books, to a corporation sole”); id., at 383 (“[Corporation sole” includes the sovereign in England).

Because I recognize that both uses of this phrase were deemed valid when § 1983 and the Dictionary Act were passed, the Court accuses me of “confusing] [the] precise definition of [this] phrase with its use ‘in a rather loose way,’ ” “to refer to the state (as opposed to a State).” Ante, at 70, n. 9, quoting Black, supra, at 143. It had never occurred to me, however, that only “precise” definitions counted as valid ones. Where the question we face is what meaning Congress attached to a particular word or phrase, we usually — and properly — are loath to conclude that Congress meant to use the word or phrase in a hypertechnical sense unless it said so. Nor does the Court’s distinction between “the state” and “a State” have any force. The suggestion, I take it, is that the phrase “bodies politic and corporate” refers only to nations rather than to the states within a nation; but then the Court must explain why so many of the sources I have quoted refer to states in addition to nations. In an opinion so utterly devoted to the rights of the States as sovereigns, moreover, it is surprising indeed to find the Court distinguishing between our sovereign States and our sovereign Nation.

In deciding what the phrase “bodies politic and corporate” means, furthermore, I do not see the relevance of the meaning of the term “public corporation.” See ante, at 69-70, n. 9. That is not the phrase chosen by Congress in the Dictionary Act, and the Court’s suggestion that this phrase is coterminous with the phrase “bodies politic and corporate” begs the question whether the latter one includes the States. Nor do I grasp the significance of this Court’s decision in United States v. Fox, 94 U. S. 315 (1877), in which the question was whether the State of New York, by including “persons” and “corporations” within the class of those to whom land could be devised, had intended to authorize devises to the United States. Ante, at 69-70, n. 9. Noting that “[t]he question is to be determined by the laws of [New York],” the *81Court held that it would require “an express definition” to hold that the word “persons” included the Federal Government, and that under state law the term “corporations” applied only to corporations created under the laws of New York. 94 U. S., at 320-321. The pertinence of these state-law questions to the issue before us today escapes me. Not only do we confront an entirely different, federal statute, but we also have an express statement, in the Dictionary Act, that the word “person” in §1 includes “bodies politic and corporate.” See also Pfizer Inc. v. India, 434 U. S., at 315, n. 15.

The relevance of the fact that § 2 of the Civil Rights Act of 1866, 14 Stat. 27, — the model for § 1 of the 1871 Act — was passed before the Dictionary Act, see ante, at 69, similarly eludes me. Congress chose to use the word “person” in the 1871 Act even after it had passed the Dictionary Act, presumptively including “bodies politic and corporate” within the category of “persons.” Its decision to do so — and its failure to indicate in the 1871 Act that the Dictionary Act’s presumption was not to apply — demonstrate that Congress did indeed intend “persons” to include bodies politic and corporate. In addition, the Dictionary Act’s definition of “person” by no means dropped from the sky. Many of the authorities cited above predate both the Dictionary Act and the 1866 Act, indicating that the word “persons” in 1866 ordinarily would have been thought to include “bodies politic and corporate,” with or without the Dictionary Act.

This last point helps to explain why it is a matter of small importance that the Dictionary Act’s definition of “person” as including bodies politic and corporate was retroactively withdrawn when the federal statutes were revised in 1874. See T. Durant, Report to Joint Committee on Revision of Laws 2 (1873). Only two months after presumptively designating bodies politic and corporate as “persons,” Congress chose the word “person” for § 1 of the Civil Rights Act. For the purpose of determining Congress’ intent in using this *82term, it cannot be decisive that, three years later, it withdrew this presumption. In fact, both the majority and dissent in Monell emphasized the 1871 version of the Dictionary Act, but neither saw fit even to mention the 1874 revision of this statute. 436 U. S., at 688-689, and nn. 51, 53 (opinion for the Court); id., at 719 (Rehnquist, J., dissenting). Even in cases, moreover, where no statutory definition of the word “persons” is available, we have not hesitated to include bodies politic and corporate within that category. See Stanley v. Schwalby, 147 U. S. 508, 517 (1893) (“[T]he word ‘person’ in the statute would include [the States] as a body politic and corporate”); Ohio v. Helvering, 292 U. S. 360, 370 (1934); United States v. Shirey, 359 U. S. 255, 257, n. 2 (1959).

Thus, the question before us is whether the presumption that the word “person” in § 1 of the Civil Rights Act of 1871 included bodies politic and corporate — and hence the States — is overcome by anything in the statute’s language and history. Certainly nothing in the statutory language overrides this presumption. The statute is explicitly directed at action taken “under color of” state law, and thus supports rather than refutes the idea that the “persons” mentioned in the statute include the States. Indeed, for almost a century — until Monroe v. Pape, 365 U. S. 167 (1961) — it was unclear whether the statute applied at all to action not authorized by the State, and the enduring significance of the first cases construing the Fourteenth Amendment, pursuant to which § 1 was passed, lies in their conclusion that the prohibitions of this Amendment do not reach private action. See Civil Rights Cases, 109 U. S. 3 (1883). In such a setting, one cannot reasonably deny the significance of § 1983’s explicit focus on state action.

Unimpressed by such arguments, the Court simply asserts that reading “States” where the statute mentions “person” would be “decidedly awkward.” Ante, at 64. The Court does not describe the awkwardness that it perceives, but I take it that its objection is that the under-color-of-law *83requirement would be redundant if States were included in the statute because States necessarily act under color of state law. But § 1983 extends as well to natural persons, who do not necessarily so act; in order to ensure that they would be liable only when they did so, the statute needed the under-color-of-law requirement. The only way to remove the redundancy that the Court sees would have been to eliminate the catchall phrase “person” altogether, and separately describe each category of possible defendants and the circumstances under which they might be liable. I cannot think of a situation not involving the Eleventh Amendment, however, in which we have imposed such an unforgiving drafting requirement on Congress.

Taking the example closest to this case, we might have observed in Monell that § 1983 was clumsily written if it included municipalities, since these, too, may act only under color of state authority. Nevertheless, we held there that the statute does apply to municipalities. 436 U. S., at 690. Similarly, we have construed the statutory term “white persons” to include “‘corporations, companies, associations, firms, partnerships, societies, and joint stock companies, as well as individuals,’” see Wilson v. Omaha Tribe, 442 U. S., at 666, quoting 1 U. S. C. § 1, despite the evident awkwardness in doing so. Indeed, virtually every time we construe the word “person” to include corporate or other artificial entities that are not individual, flesh-and-blood persons, some awkwardness results. But given cases like Monell and Wilson, it is difficult to understand why mere linguistic awkwardness should control where there is good reason to accept the “awkward” reading of a statute.

The legislative history and background of the statute confirm that the presumption created by the Dictionary Act was not overridden in § 1 of the 1871 Act, and that, even without such a presumption, it is plain that “person” in the 1871 Act must include the States. I discussed in detail the legislative history of this statute in my opinion concurring in the judg*84ment in Quern v. Jordan, 440 U. S., at 357-365, and I shall not cover that ground again here. Suffice it to say that, in my view, the legislative history of this provision, though spare, demonstrates that Congress recognized and accepted the fact that the statute was directed at the States themselves. One need not believe that the statute satisfies this Court’s heightened clear-statement principle, reserved for Eleventh Amendment cases, in order to conclude that the language and legislative history of § 1983 show that the word “person” must include the States.

As to the more general historical background of §1, we too easily forget, I think, the circumstances existing in this country when the early civil rights statutes were passed. “[V]iewed against the events and passions of the time,” United States v. Price, 383 U. S. 787, 803 (1966), I have little doubt that § 1 of the Civil Rights Act of 1871 included States as “persons.” The following brief description of the Reconstruction period is illuminating:

“The Civil War had ended in April 1865. Relations between Negroes and whites were increasingly turbulent. Congress had taken control of the entire governmental process in former Confederate States. It had declared the governments in 10 ‘unreconstructed’ States to be illegal and had set up federal military administrations in their place. Congress refused to seat representatives from these States until they had adopted constitutions guaranteeing Negro suffrage, and had ratified the Fourteenth Amendment. Constitutional conventions were called in 1868. Six of the 10 States fulfilled Congress’ requirements in 1868, the other four by 1870.

“For a few years ‘radical’ Republicans dominated the governments of the Southern States and Negroes played a substantial political role. But countermeasures were swift and violent. The Ku Klux Klan was organized by southern whites in 1866 and a similar organization appeared with the romantic title of the Knights of the *85White Camellia. In 1868 a wave of murders and assaults was launched including assassinations designed to keep Negroes from the polls. The States themselves were helpless, despite the resort by some of them to extreme measures such as making it legal to hunt down and shoot any disguised man.

“Within the Congress pressures mounted in the period between the end of the war and 1870 for drastic measures. A few months after the ratification of the Thirteenth Amendment on December 6, 1865, Congress, on April 9, 1866, enacted the Civil Rights Act of 1866 .... On June 13, 1866, the Fourteenth Amendment was proposed, and it was ratified in July 1868. In February 1869 the Fifteenth Amendment was proposed, and it was ratified in February 1870. On May 31, 1870, the Enforcement Act of 1870 was enacted.” Id., at 803-805 (footnotes omitted).

This was a Congress in the midst of altering the “ ‘balance between the States and the Federal Government.’” Ante, at 65, quoting Atascadero State Hospital v. Scanlon, 473 U. S., at 242. It was fighting to save the Union, and in doing so, it transformed our federal system. It is difficult, therefore, to believe that this same Congress did not intend to include States among those who might be liable under § 1983 for the very deprivations that were threatening this Nation at that time.

Ill

To describe the breadth of the Court’s holding is to demonstrate its unwisdom. If States are not “persons” within the meaning of § 1983, then they may not be sued under that statute regardless of whether they have consented to suit. Even if, in other words, a State formally and explicitly consented to suits against it in federal or state court, no § 1983 plaintiff could proceed against it because States are not within the statute’s category of possible defendants.

*86This is indeed an exceptional holding. Not only does it depart from our suggestion in Alabama v. Pugh, 438 U. S. 781, 782 (1978), that a State could be a defendant under § 1983 if it consented to suit, see also Quern v. Jordan, supra, at 340, but it also renders ineffective the choices some States have made to permit such suits against them. See, e. g., Della Grotta v. Rhode Island, 781 F. 2d 343 (CA1 1986). I do not understand what purpose is served, what principle of federalism or comity is promoted, by refusing to give force to a State’s explicit consent to suit.

The Court appears to be driven to this peculiar result in part by its view that “in enacting § 1983, Congress did not intend to override well-established immunities or defenses under the common law.” Ante, at 67. But the question whether States are “persons” under §1983 is separate and distinct from the question whether they may assert a defense of common-law sovereign immunity. In our prior decisions involving common-law immunities, we have not held that the existence of an immunity defense excluded the relevant state actor from the category of “persons” liable under § 1983, see, e. g., Forrester v. White, 484 U. S. 219 (1988), and it is a mistake to do so today. Such an approach entrenches the effect of common-law immunity even where the immunity itself has been waived.

For my part, I would reverse the judgment below and remand for resolution of the question whether Michigan would assert common-law sovereign immunity in defense to this suit and, if so, whether that assertion of immunity would preclude the suit.

Given the suggestion in the court below that Michigan enjoys no common-law immunity for violations of its own Constitution, Smith v. Department of Public Health, 428 Mich. 540, 641-642, 410 N. W. 2d 749, 793-794 (1987) (Boyle, J., concurring), there is certainly a possibility that that court would hold that the State also lacks immunity against § 1983 suits for violations of the Federal Constitution. *87Moreover, even if that court decided that the State’s waiver of immunity did not apply to § 1983 suits, there is a substantial question whether Michigan could so discriminate between virtually identical causes of action only on the ground that one was a state suit and the other a federal one. Cf. Testa v. Katt, 330 U. S. 386 (1947); Martinez v. California, 444 U. S. 277, 283, n. 7 (1980). Finally, even if both of these questions were resolved in favor of an immunity defense, there would remain the question whether it would be reasonable to attribute to Congress an intent to allow States to decide for themselves whether to take cognizance of § 1983 suits brought against them. Cf. Martinez, supra, at 284, and n. 8; Owen v. City of Independence, 445 U. S. 622, 647-648 (1980).

Because the court below disposed of the case on the ground that States were not “persons” within the meaning of § 1983, it did not pass upon these difficult and important questions. I therefore would remand this case to the state court to resolve these questions in the first instance.

Justice Stevens,

dissenting.

Legal doctrines often flourish long after their raison d’etre has perished.1 The doctrine of sovereign immunity rests on the fictional premise that the “King can do no wrong.”2 Even though the plot to assassinate James I in 1605, the exe*88cution of Charles I in 1649, and the Colonists’ reaction to George Ill’s stamp tax made rather clear the fictional character of the doctrine’s underpinnings/ British subjects found a gracious means of compelling the King to obey the law rather than simply repudiating the doctrine itself. They held his advisers and his agents responsible.3

In our administration of § 1983, we have also relied on fictions to protect the illusion that a sovereign State, absent consent, may not be held accountable for its delicts in federal court. Under a settled course of decision, in contexts ranging from school desegregation to the provision of public *89assistance benefits to the administration of prison systems and other state facilities, we have held the States liable under § 1983 for their constitutional violations through the artifice of naming a public officer as a nominal party. Once one strips away the Eleventh Amendment overlay applied to actions in federal court, it is apparent that the Court in these cases has treated the State as the real party in interest both for the purposes of granting prospective and ancillary relief and of denying retroactive relief. When suit is brought in state court, where the Eleventh Amendment is inapplicable, it follows that the State can be named directly as a party under § 1983.

An official-capacity suit is the typical way in which we have held States responsible for their duties under federal law. Such a suit, we have explained, “ ‘generally represents] only another way of pleading an action against an entity of which an officer is an agent.’” Kentucky v. Graham, 473 U. S. 159, 165 (1985) (quoting Monell v. New York City Dept. of Social Services, 436 U. S. 658, 690, n. 55 (1978)); see also Pennhurst State School and Hospital v. Halderman, 465 U. S. 89, 101 (1984). In the peculiar Eleventh Amendment analysis we have applied to such cases, we have recognized that an official-capacity action is in reality always against the State and balanced interests to determine whether a particular type of relief is available. The Court has held that when a suit seeks equitable relief or money damages from a state officer for injuries suffered in the past, the interests in compensation and deterrence are insufficiently weighty to override the State’s sovereign immunity. See Papasan v. Attain, 478 U. S. 265, 278 (1986); Green v. Mansour, 474 U. S. 64, 68 (1985); Edelman v. Jordan, 415 U. S. 651, 668 (1974). On the other hand, although prospective relief awarded against a state officer also “implicate[s] Eleventh Amendment concerns,” Mansour, 474 U. S., at 68, the interests in “end[ing] a continuing violation of federal law,” ibid., outweigh the interests in state sovereignty and justify *90an award under § 1983 of an injunction that operates against the State’s officers or even directly against the State itself. See, e. g., Papasan, supra, at 282; Quern v. Jordan, 440 U. S. 332, 337 (1979); Milliken v. Bradley, 433 U. S. 267, 289 (1977).

In Milliken v. Bradley, supra, for example, a unanimous Court upheld a federal-court order requiring the State of Michigan to pay $5,800,000 to fund educational components in a desegregation decree “notwithstanding [its] direct and substantial impact on the state treasury.” Id., at 289 (emphasis added).4 As Justice Powell stated in his opinion concurring in the judgment, “the State [had] been adjudged a participant in the constitutional violations, and the State therefore may be ordered to participate prospectively in a remedy otherwise appropriate.” Id., at 295. Subsequent decisions have adhered to the position that equitable relief — even “a remedy that might require the expenditure of state funds,” Papasan, supra, at 282 — may be awarded to ensure future compliance by a State with a substantive federal question determination. See also Quern v. Jordan, 440 U. S., at 337.

Our treatment of States as “persons” under § 1983 is also exemplified by our decisions holding that ancillary relief, such as attorney’s fees, may be awarded directly against the State. We have explained that “liability on the merits and responsibility for fees go hand in hand; where a defendant has not been prevailed against, either because of legal immunity *91or on the merits, § 1988 does not authorize a fee award against that defendant.” Kentucky v. Graham, supra, at 165. Nonetheless, we held in Hutto v. Finney, 437 U. S. 678 (1978), a case challenging the administration of the Arkansas prison system, that a Federal District Court could award attorney’s fees directly against the State under § 1988,5 id,., at 700; see Brandon v. Holt, 469 U. S. 464, 472 (1985), and could assess attorney’s fees for bad-faith litigation under § 1983 “ ‘to be paid out of Department of Corrections funds.’” 437 U. S., at 692. In Supreme Court of Virginia v. Consumers Union of United States, Inc., 446 U. S. 719, 739 (1980), Justice White reaffirmed for a unanimous Court that an award of fees could be entered against a State or state agency, in that case a State Supreme Court, in an in-junctive action under § 1983.6 In suits commenced in state court, in which there is no independent reason to require parties to sue nominally a state officer, we have held that attor*92ney’s fees can be awarded against the State in its own name. See Maine v. Thiboutot, 448 U. S. 1, 10-11 (1980).7

The Civil Rights Act of 1871 was “intended to provide a remedy, to be broadly construed, against all forms of official violation of federally protected rights.” Monell v. New York City Dept. of Social Services, 436 U. S., at 700-701. Our holdings that a § 1983 action can be brought against state officials in their official capacity for constitutional violations properly recognize and are faithful to that profound mandate.' If prospective relief can be awarded against state officials under § 1983 and the State is the real party in interest in such suits, the State must be a “person” which can be held liable under § 1983. No other conclusion is available. Eleventh Amendment principles may limit the State’s capacity to be sued as such in federal court. See Alabama v. Pugh, 438 U. S. 781 (1978). But since those principles are not applicable to suits in state court, see Thiboutot, supra, at 9, n. 7; Nevada v. Hall, 440 U. S. 410 (1979), there is no need to resort to the fiction of an official-capacity suit and the State may and should be named directly as a defendant in a § 1983 action.

The Court concludes, however, that “a state official in his or her official capacity, when sued for injunctive relief, would be a person under § 1983,” ante, at 71, n. 10, while that same party sued in the same official capacity is not a person when the plaintiff seeks monetary relief. It cites in support of this.,, proposition cases such as Osborn v. Bank of United States, 9 Wheat. 738 (1824), in which the Court through Chief Justice Marshall held that an action against a state auditor to recover taxes illegally collected did not constitute an action against the State. This line of authority, the Court states, “would *93not have been foreign to the 19th-century Congress that enacted § 1983." Ante, at 71, n. 10.

On the Court’s supposition, the question would be whether the complaint against a state official states a claim for the type of relief sought, not whether it will have an impact on the state treasury. See, e. g., Governor of Georgia v. Madrazo, 1 Pet. 110, 124 (1828). At least for actions in state court, as to which there could be no constitutional reason to look to the effect on the State, see Edelman v. Jordan, 415 U. S. 651 (1974), the Court’s analysis would support actions for the recovery of chattel and real property against state officials both of which were well known in the 19th century. See Poindexter v. Greenhow, 114 U. S. 270 (1884); United States v. Lee, 106 U. S. 196 (1882). Although the conclusion that a state officer sued for damages in his or her official capacity is not a “person” under § 1983 would not quite follow,8 it might nonetheless be permissible to assume that the 1871 Congress did not contemplate an action for damages payable not by the officer personally but by the State.

The Court having constructed an edifice for the purposes of the Eleventh Amendment on the theory that the State is always the real party in interest in a § 1983 official-capacity action against a state officer, I would think the majority would be impelled to conclude that the State is a “person” under § 1983. As Justice Brennan has demonstrated, there is also a compelling textual argument that States are persons under § 1983. In addition, the Court’s construction draws an illogical distinction between wrongs committed by county or municipal officials on the one hand, and those committed by state officials on the other. Finally, there is no necessity to *94import into this question of statutory construction doctrine created to protect the fiction that one sovereign cannot be sued in the courts of another sovereign. Aside from all of these reasons, the Court’s holding that a State is not a person under § 1983 departs from a long line of judicial authority based on exactly that premise.

I respectfully dissent.

Justice O’Connor

delivered the opinion of the Court.

In Will v. Michigan Dept. of State Police, 491 U. S. 58 (1989), we held that state officials “acting in their official capacities” are outside the class of “persons” subject to liability *23 under Rev. Stat. § 1979, 42 U. S. C. § 1983. 491 U. S., at 71. Petitioner takes this language to mean that § 1983 does not authorize suits against state officers for damages arising from official acts. We reject this reading of Will and hold that state officials sued in their individual capacities are “persons” for purposes of § 1983.

I

In 1988, petitioner Barbara Hafer sought election to the post of auditor general of Pennsylvania. Respondents allege that during the campaign United States Attorney James West gave Hafer a list of 21 employees in the auditor general’s office who secured their jobs through payments to a former employee of the office. App. 10. They further allege that Hafer publicly promised to fire all employees on the list if elected. Ibid.

Hafer won the election. Shortly after becoming auditor general, she dismissed 18 employees, including named respondent James Melo, Jr., on the basis that they “bought” their jobs. Melo and seven other terminated employees sued Hafer and West in Federal District Court. They asserted state and federal claims, including a claim under § 1983, and sought monetary damages. Carl Gurley and the remaining respondents in this case also lost their jobs with the auditor general soon after Hafer took office. These respondents allege that Hafer discharged them because of their Democratic political affiliation and support for her opponent in the 1988 election. Id., at 28, 35, 40. They too filed suit against Hafer, seeking monetary damages and reinstatement under § 1983.

After consolidating the Melo and Gurley actions, the District Court dismissed all claims. In relevant part, the court held that the § 1983 claims against Hafer were barred because, under Will, she could not be held liable for employment decisions made in her official capacity as auditor general.

*24 The Court of Appeals for the Third Circuit reversed this portion of the District Court’s decision. 912 F. 2d 628 (1990). As to claims for reinstatement brought against Hafer in her official capacity, the court rested on our statement in Will that state officials sued for injunctive relief in their official capacities are “persons” subject to liability under § 1983. See Will, supra, at 71, n. 10. Turning to respondents’ monetary claims, the court found that six members of the Gurley group had expressly sought damages from Hafer in her personal capacity. The remaining plaintiffs “although not as explicit, signified a similar intent.” 912 F. 2d, at 636. * The court found this critical. While Hafer’s power to hire and fire derived from her position as auditor general, it said, a suit for damages based on the exercise of this authority could be brought against Hafer in her personal capacity. Because Hafer acted under color of state law, respondents could maintain a § 1983 individual-capacity suit against her.

We granted certiorari, 498 U. S..1H8 (1991), to address the question whether state officers may be held personally liable for damages under § 1983 based upon actions taken in their official capacities.

*25 II

In Kentucky v. Graham, 473 U. S. 159 (1985), the Court sought to eliminate lingering confusion about the distinction between personal- and official-capacity suits. We emphasized that official-capacity suits “‘generally represent only another way of pleading an action against an entity of which an officer is an agent.’ ” Id., at 165 (quoting Monell v. New York City Dept. of Social Services, 436 U. S. 658, 690, n. 55 (1978)). Suits against state officials in their official capacity therefore should be treated as suits against the State. 473 U. S., at 166. Indeed, when officials sued in this capacity in federal court die or leave office, their successors automatically assume their roles in the litigation. See Fed. Rule Civ. Proc. 25(d)(1); Fed. Rule App. Proc. 43(c)(1); this Court’s Rule 35.3. Because the real party in interest in an official-capacity suit is the governmental entity and not the named official, “the entity’s ‘policy or custom’ must have played a part in the violation of federal law.” Graham, supra, at 166 (quoting Monell, supra, at 694). For the same reason, the only immunities available to the defendant in an official-capacity action are those that the governmental entity possesses. 473 U. S., at 167.

Personal-capacity suits, on the other hand, seek to impose individual liability upon a government officer for actions taken under color of state law. Thus, “[o]n the merits, to establish personal liability in a § 1983 action, it is enough to show that the official, acting under color of state law, caused the deprivation of a federal right.” Id., at 166. While the plaintiff in a personal-capacity suit need not establish a connection to governmental “policy or custom,” officials sued in their personal capacities, unlike those sued in their official capacities, may assert personal immunity defenses such as objectively reasonable reliance on existing law. Id., at 166-167.

Our decision in Will v. Michigan Dept. of State Police, 491 U. S. 58 (1989), turned in part on these differences between *26 personal- and official-capacity actions. The principal issue in Will was whether States are “persons” subject to suit under § 1983. Section 1983 provides, in relevant part:

“Every person who, under color of any statute, ordinance, regulation, custom, or usage, of any State . . . subjects, or causes to be subjected, any citizen of the United States or other person within the jurisdiction thereof to the deprivation of any rights, privileges, or immunities secured by the Constitution and laws, shall be liable to the party injured . . . .”

The Court held that interpreting the words “[ejvery person” to exclude the States accorded with the most natural reading of the law, with its legislative history, and with the rule that Congress must clearly state its intention to alter “ ‘the federal balance’” when it seeks to do so. Will, supra, at 65 (quoting United States v. Bass, 404 U. S. 336, 349 (1971)).

The Court then addressed the related question whether state officials, sued for monetary relief in their official capacities, are persons under § 1983. We held that they are not. Although “state officials literally are persons,” an official-capacity suit against a state officer “is not a suit against the official but rather is a suit against the official’s office. As such it is no different from a suit against the State itself.” 491 U. S., at 71 (citation omitted).

Summarizing our holding, we said: “[NJeither a State nor its officials acting in their official capacities are ‘persons’ under § 1983.” Ibid. Hafer relies on this recapitulation for the proposition that she may not be held personally liable under § 1983 for discharging respondents because she “act[ed]” in her official capacity as auditor general of Pennsylvania. Of course, the claims considered in Will were official-capacity claims; the phrase “acting in their official capacities” is best understood as a reference to the capacity in which the state officer is sued, not the capacity in which the officer inflicts the alleged injury. To the extent that Will *27 allows the construction Hafer suggests, however, we now eliminate that ambiguity.

A

Will itself makes clear that the distinction between official-capacity suits and personal-capacity suits is more than “a mere pleading device.” Ibid. State officers sued for damages in their official capacity are not “persons” for purposes of the suit because they assume the identity of the government that employs them. Ibid. By contrast, officers sued in their personal capacity come to court as individuals. A government official in the role of personal-capacity defendant thus fits comfortably within the statutory term “person.” Cf. id., at 71, n. 10 (“[A] state official in his or her official capacity, when sued for injunctive relief, would be a person under § 1983 because ‘official-capacity actions for prospective relief are not treated as actions against the State’ ”) (quoting Graham, 473 U. S., at 167, n. 14).

Hafer seeks to overcome the distinction between official- and personal-capacity suits by arguing that §1983 liability turns not on the capacity in which state officials are sued, but on the capacity in which they acted when injuring the plaintiff. Under Will, she asserts, state officials may not be held liable in their personal capacity for actions they take in their official capacity. Although one Court of Appeals has endorsed this view, see Cowan v. University of Louisville School of Medicine, 900 F. 2d 936, 942-943 (CA6 1990), we find it both unpersuasive as an interpretation of § 1983 and foreclosed by our prior decisions.

Through § 1983, Congress sought “to give a remedy to parties deprived of constitutional rights, privileges and immunities by an official’s abuse of his position.” Monroe v. Pape, 365 U. S. 167, 172 (1961). Accordingly, it authorized suits to redress deprivations of civil rights by persons acting “under color of any [state] statute, ordinance, regulation, custom, or usage.” 42 U. S. C. § 1983. The requirement of action under color of state law means that Hafer may be liable for *28 discharging respondents precisely because of her authority as auditor general. We cannot accept the novel proposition that this same official authority insulates Hafer from suit.

In an effort to limit the scope of her argument, Hafer distinguishes between two categories of acts taken under color of state law: those outside the official’s authority or not essential to the operation of state government, and those both within the official’s authority and necessary to the performance of governmental functions. Only the former group, she asserts, can subject state officials to personal liability under § 1983; the latter group (including the employment decisions at issue in this case) should be considered acts of the State that cannot give rise to a personal-capacity action.

The distinction Hafer urges finds no support in the broad language of § 1983. To the contrary, it ignores our holding that Congress enacted § 1983 “ ‘to enforce provisions of the Fourteenth Amendment against those who carry a badge of authority of a State and represent it in some capacity, whether they act in accordance with their authority or misuse it.’ ” Scheuer v. Rhodes, 416 U. S. 232, 243 (1974) (quoting Monroe v. Pape, supra, at 171-172). Because of that intent, we have held that in § 1983 actions the statutory requirement of action “under color of” state law is just as broad as the Fourteenth Amendment’s “state action” requirement. Lugar v. Edmondson Oil Co., 457 U. S. 922, 929 (1982).

Furthermore, Hafer’s distinction cannot be reconciled with our decisions regarding immunity of government officers otherwise personally liable for acts done in the course of their official duties. Her theory would absolutely immunize state officials from personal liability for acts within their authority and necessary to fulfilling governmental responsibilities. Yet our cases do not extend absolute immunity to all officers who engage in necessary official acts. Rather, immunity from suit under § 1983 is “predicated upon a considered inquiry into the immunity historically accorded the relevant *29 official at common law and the interests behind it,” Imbler v. Pachtman, 424 U. S. 409, 421 (1976), and officials seeking absolute immunity must show that such immunity is justified for the governmental function at issue, Burns v. Reed, 500 U. S. 478, 486-487 (1991).

This Court has refused to extend absolute immunity beyond a very limited class of officials, including the President of the United States, legislators carrying out their legislative functions, and judges carrying out their judicial functions, “whose special functions or constitutional status requires complete protection from suit.” Harlow v. Fitzgerald, 457 U. S. 800, 807 (1982). State executive officials are not entitled to absolute immunity for their official actions. Scheuer v. Rhodes, supra. In several instances, moreover, we have concluded that no more than a qualified immunity attaches to administrative employment decisions, even if the same official has absolute immunity when performing other functions. See Forrester v. White, 484 U. S. 219 (1988) (dismissal of court employee by state judge); Harlow v. Fitzgerald, supra (discharge of Air Force employee, allegedly orchestrated by senior White House aides) (action under Bivens v. Six Unknown Fed. Narcotics Agents, 403 U. S. 388 (1971)); Davis v. Passman, 442 U. S. 228 (1979) (dismissal of congressional aide) (Bivens action). That Hafer may assert personal immunity within the framework of these cases in no way supports her argument here.

B

Hafer further asks us to read Will’s language concerning suits against state officials as establishing the limits of liability under the Eleventh Amendment. She asserts that imposing personal liability on officeholders may infringe on state sovereignty by rendering government less effective; thus, she argues, the Eleventh Amendment forbids personal-capacity suits against state officials in federal court.

*30 Most certainly, Will’s holding does not rest directly on the Eleventh Amendment. Whereas the Eleventh Amendment bars suits in federal court “by private parties seeking to impose a liability which must be paid from public funds in the state treasury,” Edelman v. Jordan, 415 U. S. 651, 663 (1974), Will arose from a suit in state court. We considered the Eleventh Amendment in Will only because the fact that Congress did not intend to override state immunity when it enacted § 1983 was relevant to statutory construction: “Given that a principal purpose behind the enactment of § 1983 was to provide a federal forum for civil rights claims,” Congress’ failure to authorize suits against States in federal courts suggested that it also did not intend to authorize such claims in state courts. 491 U. S., at 66.

To the extent that Hafer argues from the Eleventh Amendment itself, she makes a claim that failed in Scheuer v. Rhodes, supra. In Scheuer, personal representatives of the estates of three students who died at Kent State University in May 1970 sought damages from the Governor of Ohio and other state officials. The District Court dismissed their complaints on the theory that the suits, although brought against state officials in their personal capacities, were in substance actions against the State of Ohio and therefore barred by the Eleventh Amendment.

We rejected this view. “[S]ince Ex parte Young, 209 U. S. 123 (1908),” we said, “it has been settled that the Eleventh Amendment provides no shield for a state official confronted by a claim that he had deprived another of a federal right under the color of state law.” Scheuer, supra, at 237. While the doctrine of Ex parte Young does not apply where a plaintiff seeks damages from the public treasury, damages awards against individual defendants in federal courts “are a permissible remedy in some circumstances notwithstanding the fact that they hold public office.” 416 U. S., at 238. That is, the Eleventh Amendment does not erect a barrier *31 against suits to impose “individual and personal liability” on state officials under § 1983. Ibid.

To be sure, imposing personal liability on state officers may hamper their performance of public duties. But such concerns are properly addressed within the framework of our personal immunity jurisprudence. See Forrester v. White, supra, at 223. Insofar as respondents seek damages against Hafer personally, the Eleventh Amendment does not restrict their ability to sue in federal court.

We hold that state officials, sued in their individual capacities, are “persons” within the meaning of § 1983. The Eleventh Amendment does not bar such suits, nor are state officers absolutely immune from personal liability under § 1983 solely by virtue of the “official” nature of their acts.

The judgment of the Court of Appeals is

Affirmed.

Justice Thomas took no part in the consideration or decision of this case.

Mr. Justice Brennan

delivered the opinion of the Court.

Petitioners, a class of female employees of the Department of Social Services and of the Board of Education of the city of New York, commenced this action under 42 U. S. C. § 1983 in July 1971.1 The gravamen of the complaint was that the *661Board and the Department had as a matter of official policy compelled pregnant employees to take unpaid leaves of absence before such leaves were required for medical reasons.2 Cf. Cleveland Board of Education v. LaFleur, 414 U. S. 632 (1974). The suit sought injunctive relief and backpay for periods of unlawful forced leave. Named as defendants in the action were the Department and its Commissioner, the Board and its Chancellor, and the city of New York and its Mayor. In each case, the individual defendants were sued solely in their official capacities.3

On cross-motions for summary judgment, the District Court for the Southern District of New York held moot petitioners’ claims for injunctive and declaratory relief since the city of New York and the Board, after the filing of the complaint, had changed their policies relating to maternity leaves so that no pregnant employee would have to take leave unless she was medically unable to continue to perform her job. 394 F. Supp. 853, 855 (1975). No one now challenges this conelu*662sion. The court did conclude, however, that the acts complained of were unconstitutional under LaFleur, supra. 394 F. Supp., at 855. Nonetheless plaintiffs’ prayers for backpay were denied because any such damages would come ultimately from the city of New York and, therefore, to hold otherwise would be to “circumvenft]” the immunity conferred on municipalities by Monroe v. Pape, 365 U. S. 167 (1961). See 394 F. Supp., at 855.

On appeal, petitioners renewed their arguments that the Board of Education4 was not a “municipality” within the meaning of Monroe v. Pape, supra, and that, in any event, the District Court had erred in barring a damages award against the individual defendants. The Court of Appeals for the Second Circuit rejected both contentions. The court first held that the Board of Education was not a “person” under § 1983 because “it performs a vital governmental function . . ., and, significantly, while it has the right to determine how the funds appropriated to it shall be spent ... , it has no final say in deciding what its appropriations shall be.” 532 F. 2d 259, 263 (1976). The individual defendants, however, were “persons” under § 1983, even when sued solely in their official capacities. 532 F. 2d, at 264. Yet, because a damages award would “have to be paid by a city that was held not to be amenable to such an action in Monroe v. Pape," a damages action against officials sued in their official capacities could not proceed. Id., at 265.

We granted certiorari in this case, 429 U. S. 1071, to consider

“Whether local governmental officials and/or local independent school boards are 'persons’ within the meaning of 42 U. S. C. § 1983 when equitable relief in the nature of back pay is sought against them in their official capacities?” Pet. for Cert. 8.

*663Although, after plenary consideration, we have decided the merits of over a score of cases brought under § 1983 in which the principal defendant was a school board5 — and, indeed, in some of which § 1983 and its jurisdictional counterpart, 28 U. S. C. § 1343, provided the only basis for jurisdiction6 — we indicated in Mt. Healthy City Board of Education v. Doyle, 429 U. S. 274, 279 (1977), last Term that the question presented here was open and would be decided “another day.” That other day has come and we now overrule Monroe v. Pape, supra, insofar as it holds that local governments are wholly immune from suit under § 1983.7

*664I

In Monroe v. Pape, we held that “Congress did not undertake to bring municipal corporations within the ambit of [§ 1983].” 365 U. S., at 187. The sole basis for this conclusion was an inference drawn from Congress’ rejection of the “Sherman amendment” to the bill which became the Civil Rights Act of 1871, 17 Stat. 13, the precursor of § 1983. The amendment would have held a municipal corporation liable for damage done to the person or property of its inhabitants by private persons “riotously and tumultuously assembled.”8 Cong. Globe, 42d Cong., 1st Sess., 749 (1871) (hereinafter Globe). Although the Sherman amendment did not seek to amend § 1 of the Act, which is now § 1983, and although the nature of the obligation created by that amendment was vastly different from that created by § 1, the Court nonetheless concluded in Monroe that Congress must have meant to exclude municipal corporations from the coverage of § 1 because “ 'the House [in voting against the Sherman amendment] had solemnly decided that in their judgment Congress had no constitutional power to impose any obligation upon county and town organizations, the mere instrumentality for the administration of state law.’ ” 365 U. S., at 190 (emphasis added), quoting Globe 804 (Rep. Poland). This statement, we thought, showed that Congress doubted its “constitutional power . . . to impose civil liability on municipalities,” 365 U. S., at 190 (emphasis added), and that such doubt would have extended to any type of civil liability.9

*665A fresh analysis of the debate on the Civil Rights Act of 1871, and particularly of the case law which each side mustered in its support, shows, however, that Monroe incorrectly equated the “obligation” of which Representative Poland spoke with “civil liability.”

A. An Overview

There are three distinct stages in the legislative consideration of the bill which became the Civil Rights Act of 1871. On March 28, 1871, Representative Shellabarger, acting for a House select committee, reported H. R. 320, a bill “to enforce the provisions of the fourteenth amendment to the Constitution of the United States, and for other purposes.” H. R. 320 contained four sections. Section 1, now codified as 42 U. S. C. § 1983, was the subject of only limited debate and was passed without amendment.10 Sections 2 through 4 dealt primarily with the “other purpose” of suppressing Ku Klux Klan violence in the Southern States.11 The wisdom and constitutionality of these sections — not § 1, now § 1983 — were the subject of almost all congressional debate and each of these sections was amended. The House finished its initial debates on H. R. 320 on April 7, 1871, and one week later the Senate also voted out a bill.12 Again, debate on § 1 of the bill was limited and that section was passed as introduced.

*666Immediately prior to the vote on H. R. 320 in the Senate, Senator Sherman introduced his amendment.13 This was not an amendment to § 1 of the bill, but was to be added as § 7 at the end of the bill. Under the Senate rules, no discussion of the amendment was allowed and, although attempts were made to amend the amendment, it was passed as introduced. In this form, the amendment did not place liability on municipal corporations, but made any inhabitant of a municipality liable for damage inflicted by persons “riotously and tumultuously assembled.” 14

The House refused to acquiesce in a number of amendments made by the Senate, including the Sherman amendment, and the respective versions of H. R. 320 were therefore sent to a conference committee. Section 1 of the bill, however, was not a subject of this conference since, as noted, it was passed verbatim as introduced in both Houses of Congress.

On April 18, 1871, the first conference committee completed its work on H. R. 320. The main features of the conference committee draft of the Sherman amendment were these: 15 First, a cause of action was given to persons injured by

“any persons riotously and tumultuously assembled together . . . with intent to deprive any person of any right conferred upon him by the Constitution and laws of the United States, or to deter him or punish him for exercising such right, or by reason of his race, color, or previous condition of servitude . . . .”

*667Second, the bill provided that the action would, be against the county, city, or parish in which the riot had occurred and that it could be maintained by either the person injured or his legal representative. Third, unlike the amendment as proposed, the conference substitute made the government defendant liable on the judgment if it was not satisfied against individual defendants who had committed the violence. If a municipality were liable, the judgment against it could be collected

“by execution, attachment, mandamus, garnishment, or any other proceeding in aid of execution or applicable to the enforcement of judgments against municipal corporations; and such judgment [would become] a lien as well upon all moneys in the treasury of such county, city, or parish, as upon the other property thereof.”

In the ensuing debate on the first conference report, which was the first debate of any kind on the Sherman amendment, Senator Sherman explained that the purpose of his amendment was to enlist the aid of persons of property in the enforcement of the civil rights laws by making their property “responsible” for Ku Klux Klan damage.16 Statutes drafted on a similar theory, he stated, had long been in force in England and were in force in 1871 in a number of States.17 *668Nonetheless there were critical differences between the conference substitute and extant state and English statutes: The conference substitute, unlike most state riot statutes, lacked a short statute of limitations and imposed liability on the government defendant whether or not it had notice of the impending riot, whether or not the municipality was authorized to exercise a police power, whether or not it exerted all reasonable efforts to stop the riot, and whether or not the rioters were caught and punished.18

The first conference substitute passed the Senate but was rejected by the House. House opponents, within whose ranks were some who had supported § 1, thought the Federal Government could not, consistent with the Constitution, obligate municipal corporations to keep the peace if those corporations were neither so obligated nor so authorized by their state charters. And, because of this constitutional objection, opponents of the Sherman amendment were unwilling to impose damages liability for nonperformance of a duty which Congress could not require municipalities to perform. This position is reflected in Representative Poland's statement that is quoted in Monroe.19

Because the House rejected the first conference report a second conference was called and it duly issued its report. The second conference substitute for the Sherman amendment abandoned municipal liability and, instead, made “any per*669son or persons having knowledge [that a conspiracy to violate civil rights was afoot], and having power to prevent or aid in preventing the same,” who did not attempt to stop the same, liable to any person injured by the conspiracy.20 The amendment in this form was adopted by both Houses of Congress and is now codified as 42 U. S. C. § 1986.

The meaning of the legislative history sketched above can most readily be developed by first considering the debate on the report of the first conference committee. This debate shows conclusively that the constitutional objections raised against the Sherman amendment — on which our holding in Monroe was based, see supra, at 664 — would not have prohibited congressional creation of a civil remedy against state municipal corporations that infringed federal rights. Because § 1 of the Civil Rights Act does not state expressly that municipal corporations come within its ambit, it is finally necessary to interpret § 1 to confirm that such corporations were indeed intended to be included within the “persons” to whom that section applies.

B. Debate on the First Conference Report

The style of argument adopted by both proponents and opponents of the Sherman amendment in both Houses of Congress was largely legal, with frequent references to cases decided by this Court and the Supreme Courts of the several States. Proponents of the Sherman amendment did not, however, discuss in detail the argument in favor of its constitutionality. Nonetheless, it is possible to piece together such an argument from the debates on the first conference report and those on § 2 of the civil rights bill, which, because it allowed the Federal Government to prosecute crimes “in the States,” had also raised questions of federal power. The account of Representative Shellabarger, the House sponsor of H. R. 320, is the most complete.

*670Shellabarger began his discussion of H. R. 320 by stating that “there is a domain of constitutional law involved in the right consideration of this measure which is wholly unexplored.” Globe App. 67. There were analogies, however. With respect to the meaning of § 1 of the Fourteenth Amendment, and particularly its Privileges or Immunities Clause, Shellabarger relied on the statement of Mr. Justice Washington in Corfield v. Coryell, 4 Wash. C. C. 371 (CC ED Pa. 1825), which defined the privileges protected by Art. IV:

“ What these fundamental privileges are[,] it would perhaps be more tedious than difficult to enumerate. They may, however, be all comprehended under the following general heads: protection by the Government;’ — •

“Mark that—

“ ‘protection by the Government; the enjoyment of life and liberty, with the right to acquire and possess property of every kind, and to pursue and obtain happiness and safety ....’” Globe App. 69 (emphasis added), quoting 4 Wash. C. C., at 380-381.

Building on his conclusion that citizens were owed protection — a conclusion not disputed by opponents of the Sherman amendment21 — Shellabarger then considered Congress’ role in providing that protection. Here again there were precedents:

“[Congress has always] assumed to enforce, as against *671the States, and also persons, every one of the provisions of the Constitution. Most of the provisions of the Constitution which restrain and directly relate to the States, such as those in [Art. I, § 10,] relate to the divisions of the political powers of the State and General Governments. . . . These prohibitions upon political powers of the States are all of such nature that they can be, and even have been, . . . enforced by the courts of the United States declaring void all State acts of encroachment on Federal powers. Thus, and thus sufficiently, has the United States 'enforced’ these provisions of the Constitution. But there are some that are not of this class. These are where the court secures the rights or the liabilities of persons within the States, as between such persons and the States.

"These three are: first, that as to fugitives from justice; [22] second, that as to fugitives from service, (or slaves;) [23] third, that declaring that the 'citizens of each State shall be entitled to all the privileges and immunities of citizens in the several States.’[24]

*672“And, sir, every one of these' — the only provisions where it was deemed that legislation was required to enforce the constitutional provisions — -the only three where the rights or liabilities of persons in the States, as between these persons and the States, are directly provided for, Congress has by legislation affirmatively interfered to protect . . . such persons.” Globe App. 69-70.

Of legislation mentioned by Shellabarger, the closest analog of the Sherman amendment, ironically, was the statute implementing the fugitives from justice and fugitive slave provisions of Art. IV — the Act of Feb. 12, 1793, 1 Stat. 302 — the constitutionality of which had been sustained in 1842, in Prigg v. Pennsylvania, 16 Pet. 539. There, Mr. Justice Story, writing for the Court, held that Art. IV gave slaveowners a federal right to the unhindered possession of their slaves in whatever State such slaves might be found. 16 Pet., at 612. Because state process for recovering runaway slaves might be inadequate or even hostile to the rights of the slaveowner, the right intended to be conferred could be negated if left to state implementation. Id., at 614. Thus, since the Constitution guaranteed the right and this in turn required a remedy, Story held it to be a “natural inference” that Congress had the power itself to ensure an appropriate (in the Necessary and Proper Clause sense) remedy for the right. Id., at 615.

Building on Prigg, Shellabarger argued that a remedy against municipalities and counties was an appropriate — and hence constitutional — method for ensuring the protection which the Fourteenth Amendment made every citizen’s federal right.25 This much was clear from the adoption of such statutes by the several States as devices for suppressing riot.26 Thus, said Shellabarger, the only serious question remaining *673was “whether, since a county is an integer or part of a State, the United States can impose upon it, as such, any obligations to keep the peace in obedience to United States laws.”27 This he answered affirmatively, citing Board of Comm’rs v. Aspinwall, 24 How. 376 (1861), the first of many cases28 upholding the power of federal courts to enforce the Contract Clause against municipalities.29

House opponents of the Sherman amendment — whose views are particularly important since only the House voted down the amendment — did not dispute Shellabarger’s claim that the Fourteenth Amendment created a federal right to protection, see n. 21, supra, but they argued that the local units of government upon which the amendment fastened liability were not obligated to keep the peace at state law and further that the Federal Government could not constitutionally require local governments to create police forces, whether this requirement was levied directly, or indirectly by imposing damages for breach of the peace on municipalities. The most complete statement of this position is that of Representative Blair: 30

“The proposition known as the Sherman amend*674ment ... is entirely new. It is altogether without a precedent in this country. . . . That amendment claims the power in the General Government to go into the States of this Union and lay such obligations as it may please upon the municipalities, which are the creations of the States alone. . . .

“. . . [H]ere it is proposed, not to carry into effect an obligation which rests upon the municipality, but to *675create that obligation, and that is the provision I am unable to assent to. The parallel of the hundred does not in the least meet the case. The power that laid the obligation upon the hundred first put the duty upon the hundred that it should perform in that regard, and failing to meet the obligation which had been laid upon it, it was very proper that it should suffer damage for its neglect....

. . [Tjhere are certain rights and duties that belong to the States, . . . there are certain powers that inhere in the State governments. They create these municipalities, they say what their powers shall be and what their obligations shall be. If the Government of the United States can step in and add to those obligations, may it not utterly destroy the municipality? If it can say that it shall be liable for damages occurring from a riot, . . . where [will] its power . . . stop and what obligations . . . might [it] not lay upon a municipality. . . .

“Now, only the other day, the Supreme Court . . . decided [in Collector v. Day, 11 Wall. 113 (1871)] that there is no power in the Government of the United States, under its authority to tax, to tax the salary of a State officer. Why? Simply because the power to tax involves the power to destroy, and it was not the intent to give the Government of the United States power to destroy the government of the States in any respect. It was held also in the case of Prigg vs. Pennsylvania [16 Pet. 539 (1842)] that it is not within the power of the Congress of the United States to lay duties upon a State officer; that we cannot command a State officer to do any duty whatever, as such; and I ask . . . the difference between that and commanding a municipality, which is equally the creature of the State, to perform a duty.” Globe 795.

Any attempt to impute a unitary constitutional theory to opponents of the Sherman amendment is, of course, fraught *676with difficulties, not the least of which is that most Members of Congress did not speak to the issue of the constitutionality of the amendment. Nonetheless, two considerations lead us to conclude that opponents of the Sherman amendment found it unconstitutional substantially because of the reasons stated by Representative Blair: First, Blair’s analysis is precisely that of Poland, whose views were quoted as authoritative in Monroe, see supra, at 664, and that analysis was shared in large part by all House opponents who addressed the constitutionality of the Sherman amendment.31 Second, Blair’s exegesis of the reigning constitutional theory of his day, as we shall explain, was clearly supported by precedent — albeit precedent that has not survived, see Ex parte Virginia, 100 U. S. 339, 347-348 (1880); Graves v. New York ex rel. O’Keefe, 306 U. S. 466, 486 (1939) — and no other constitutional formula was advanced by participants in the House debates.

Collector v. Day, cited by Blair, was the clearest and, at the time of the debates, the most recent pronouncement of a doctrine of coordinate sovereignty that, as Blair stated, placed limits on even the enumerated powers of the National Government in favor of protecting state prerogatives. There, the Court held that the United States could not tax the income of Day, a Massachusetts state judge, because the independence of the States within their legitimate spheres would be imperiled if the instrumentalities through which States executed their powers were "subject to the control of another and distinct government.” 11 Wall., at 127. Although the Court in Day apparently rested this holding in part on the proposition that the taxing "power acknowledges no limits but the will of the legislative body imposing the tax,” id., at 125-126; cf. McCulloch v. Maryland, 4 Wheat. 316 (1819), the Court had in other cases limited other national powers in order to avoid interference with the States.32

*677In Prigg v. Pennsylvania, for example, Mr. Justice Story, in addition to confirming a broad national power to legislate under the Fugitive Slave Clause, see supra, at 672, held that Congress could not “insist that states . . . provide means to carry into effect the duties of the national government.” 16 Pet., at 615-616.83 And Mr. Justice McLean agreed that, “[a]s a general principle,” it was true “that Congress had no power to impose duties on state officers, as provided in the [Act of Feb. 12, 1793].” Nonetheless he wondered whether Congress might not impose “positive” duties on state officers where a clause of the Constitution, like the Fugitive Slave Clause, seemed to require affirmative government assistance, rather than restraint of government, to secure federal rights. See id., at 664-665.

Had Mr. Justice McLean been correct in his suggestion that, where the Constitution envisioned affirmative government assistance, the States or their officers or instrumentalities could be required to provide it, there would have been little doubt that Congress could have insisted that municipalities afford by “positive” action the protection34 owed individuals under § 1 of the Fourteenth Amendment whether or not municipalities were obligated by state law to keep' the peace. However, any such argument, largely foreclosed by Prigg, was made *678impossible by the Court’s holding in Kentucky v. Dennison, 24 How. 66 (1861). There, the Court was asked to require Dennison, the Governor of Ohio, to hand over Lago, a fugitive from justice wanted in Kentucky, as required by § 1 of the Act of Feb. 12, 1793,35 which implemented Art. IV, § 2, cl. 2, of the Constitution. Mr. Chief Justice Taney, writing for a unanimous Court, refused to enforce that section of the Act:

“[W]e think it clear, that the Federal Government, under the Constitution, has no power to impose on a State officer, as such, any duty whatever, and compel him to perform it; for if it possessed this power, it might overload the officer with duties which would fill up' all his time, and disable him from performing his obligations to the State, and might impose on him duties of a character incompatible with the rank and dignity to which he was elevated by the State.” 24 How., at 107-108.

The rationale of Dennison — that the Nation could not impose duties on state officers since that might impede States in their legitimate activities — is obviously identical to that which animated the decision in Collector v. Day. See supra, at 676. And, as Blair indicated, municipalities as instrumen-talities through which States executed their policies could be equally disabled from carrying out state policies if they were also obligated to carry out federally imposed duties. Although no one cited Dennison by name, the principle for which it *679stands was well known to Members of Congress,36 many of whom discussed Day 37 as well as a series of State Supreme Court cases38 in the mid-1860’s which had invalidated a federal tax on the process of state courts on the ground that the tax threatened the independence of a vital state function.39 Thus, there was ample support for Blair’s view that the Sherman amendment, by putting municipalities to the Hobson’s choice of keeping the peace or paying civil damages, attempted to impose obligations on municipalities by indirection that could not be imposed directly, thereby threatening to “destroy the government of the States.” Globe 795.

If municipal liability under § 1 of the Civil Rights Act of 1871 created a similar Hobson’s choice, we might conclude, as Monroe did, that Congress could not have intended municipalities to be among the “persons” to which that section applied. But this is not the case.

Hirst, opponents expressly distinguished between imposing an obligation to keep the peace and merely imposing civil liability for damages on a municipality that was obligated by state law to keep the peace, but which had not in violation of the Fourteenth Amendment. Representative Poland, for example, reasoning from Contract Clause precedents, indicated that Congress could constitutionally confer jurisdiction on the federal courts to entertain suits seeking to hold municipalities *680liable for using their authorized powers in violation of the Constitution — which is as far as § 1 of the Civil Nights Act went:

“I presume . . . that where a State had imposed a duty [to keep the peace] upon [a] municipality ... an action would be allowed to be maintained against them in the courts of the United States under the ordinary restrictions as to jurisdiction. But the enforcing a liability, existing by their own contract, or by a State law, in the courts, is a very widely different thing from devolving a new duty or liability upon them by the national Government, which has no power either to create or destroy them, and no power or control over them whatever.” Globe 794.

Representative Burchard agreed:

“[Tjhere is no duty imposed by the Constitution of the United States, or usually by State laws, upon a county to protect the people of that county against the commission of the offenses herein enumerated, such as the burning of buildings or any other injury to property or injury to person. Police powers are not conferred upon counties as corporations; they are conferred upon cities that have qualified legislative power. And so far as cities are concerned, where the equal protection required to be afforded by a State is imposed upon a city by State laws, perhaps the United States courts could enforce its performance. But counties ... do not have any control of the police . . . .” Id., at 795.

See also the views of Rep. Willard, discussed at n. 30, supra.

Second, the doctrine of dual sovereignty apparently put no limit on the power of federal courts to enforce the Constitution against municipalities that violated it. Under the theory of dual sovereignty set out in Prigg, this is quite understandable. So long as federal courts were vindicating the Federal Constitution, they were providing the “positive” government action *681required to protect federal constitutional rights and no question was raised of enlisting the States in “positive” action. The limits of the principles announced in Dennison and Day are not so well defined in logic, but are clear as a matter of history. It must be remembered that the same Court which rendered Day also vigorously enforced the Contract Clause against municipalities — an enforcement effort which included various forms of “positive” relief, such as ordering that taxes be levied and collected to discharge federal-court judgments, once a constitutional infraction was found.40 Thus, federal judicial enforcement of the Constitution's express limits on state power, since it was done so frequently, must, notwithstanding anything said in Dennison or Day, have been permissible, at least so long as the interpretation of the Constitution was left in the hands of the judiciary. Since § 1 of the Civil Rights Act simply conferred jurisdiction on the federal courts to enforce § 1 of the Fourteenth Amendment — a situation precisely analogous to the grant of diversity jurisdiction under which the Contract Clause was enforced against munici*682palities — there is no reason to suppose that opponents of the Sherman amendment would have found any constitutional barrier to § 1 suits against municipalities.

Finally, the very votes of those Members of Congress, who opposed the Sherman amendment but who had voted for § 1, confirm that the liability imposed by § 1 was something very different from that imposed by the amendment. Section 1 without question could be used to obtain a damages judgment against state or municipal officials who violated federal constitutional rights while acting under color of law.41 However, for Prigg-Dennison-Day purposes, as Blair and others recognized,42 there was no distinction of constitutional magnitude between officers and agents — including corporate agents — of the State: Both were state instrumentalities and the State could be impeded no matter over which sort of instrumentality the Federal Government sought to assert its power. Dennison and Day, after all, were not suits against municipalities but against officers, and Blair was quite conscious that he was extending these cases by applying them to municipal corporations.43 Nonetheless, Senator Thurman, who gave the most exhaustive critique of § 1 — inter alia, complaining that it would be applied to state officers, see Globe App. 217 — and who opposed both § 1 and the Sherman amendment, the latter on Prigg grounds, agreed unequivocally that § 1 was constitu*683tional.44 Those who voted for § 1 must similarly have believed in its constitutionality despite Prigg, Dennison, and Day.

C. Debate on § 1 of the Civil Rights Bill

From the foregoing discussion, it is readily apparent that nothing said in debate on the Sherman amendment would have prevented holding a municipality liable under § 1 of the Civil Rights Act for its own violations of the Fourteenth Amendment. The question remains, however, whether the general language describing those to be liable under § 1 — “any person” — covers more than natural persons. An examination of the debate on § 1 and application of appropriate rules of construction show unequivocally that § 1 was intended to cover legal as well as natural persons.

Representative Shellabarger was the first to explain the function of § 1:

“[Section 1] not only provides a civil remedy for persons whose former condition may have been that of slaves, but also to all people where, under color of State law, they or any of them may be deprived of rights to which they are entitled under the Constitution by reason and virtue of their national citizenship.” Globe App. 68.

By extending a remedy to all people, including whites, § 1 went beyond the mischief to which the remaining sections of the 1871 Act were addressed. Representative Shellabarger also stated without reservation that the constitutionality of § 2 of the Civil Rights Act of 1866 controlled the constitutionality of § 1 of the 1871 Act, and that the former had been *684approved by “the supreme courts of at least three States of this Union” and by Mr. Justice Swayne, sitting on circuit, who had concluded: “'We have no doubt of the constitutionality of every provision of this act/ ” Globe App. 68. Representative Shellabarger then went on to describe how the courts would and should interpret § 1:

“This act is remedial, and in aid of the preservation of human liberty and human rights. All statutes and constitutional provisions authorizing such statutes are liberally and beneficently construed. It would be most strange and, in civilized law, monstrous were this not the rule of interpretation. As has been again and again decided by your own Supreme Court of the United States, and everywhere else where there is wise judicial interpretation, the largest latitude consistent with the words employed is uniformly given in construing such statutes and constitutional provisions as are meant to protect and defend and give remedies for their wrongs to all the people. . . . Chief Justice Jay and also Story say:

“ 'Where a power is remedial in its nature there is much reason to contend that it ought to be construed liberally, and it is generally adopted in the interpretation of laws.’ — 1 Story on Constitution, sec. 429.” Globe App., at 68.

The sentiments expressed in Representative Shellabarger’s opening speech were echoed by Senator Edmunds, the manager of H. R. 320 in the Senate:

“The first section is one that I believe nobody objects to, as defining the rights secured by the Constitution of the United States when they are assailed by any State law or under color of any State law, and it is merely carrying out the principles of the civil rights bill [of 1866], which have since become a part of the Constitution.” Globe 568.

*685“[Section 1 is] so very simple and really reenact[s] the Constitution.” Id., at 569.

And he agreed that the bill “secure[d] the rights of white men as much as of colored men.” Id., at 696.

In both Houses, statements of the supporters of § 1 corroborated that Congress, in enacting § 1, intended to give a broad remedy for violations of federally protected civil rights.45 Moreover, since municipalities through their official *686acts could, equally with natural persons, create the harms intended to be remedied by § 1, and, further, since Congress intended § 1 to be broadly construed, there is no reason to suppose that municipal corporations would have been excluded from the sweep of § 1. Cf., e. g., Ex parte Virginia, 100 U. S. 339, 346-347 (1880); Home Tel. & Tel. Co. v. Los Angeles, 227 U. S. 278, 286-287, 294-296 (1913). One need not rely on this inference alone, however, for the debates show that Members of Congress understood “persons” to include municipal corporations.

Representative Bingham, for example, in discussing § 1 of the bill, explained that he had drafted § 1 of the Fourteenth Amendment with the case of Barron v. Mayor of Baltimore, 7 Pet. 243 (1833), especially in mind. “In [that] case the *687city had taken private property for public use, without compensation . . . , and there was no redress for the wrong Globe App. 84 (emphasis added). Bingham’s further remarks clearly indicate his view that such takings by cities, as had occurred in Barron, would be redressable under § 1 of the bill. See Globe App. 85. More generally, and as Bingham’s remarks confirm, § 1 of the bill would logically be the vehicle by which Congress provided redress for takings, since that section provided the only civil remedy for Fourteenth Amendment violations and that Amendment unequivocally prohibited uncompensated takings.46 Given this purpose, it beggars reason to suppose that Congress would have exempted municipalities from suit, insisting instead that compensation for a taking come from an officer in his individual capacity rather than from the government unit that had the benefit of the property taken.47

In addition, by 1871, it was well understood that corporations should be treated as natural persons for virtually all purposes of constitutional and statutory analysis. This had not always been so. When this Court first considered the question of the status of corporations, Mr. Chief Justice Marshall, writing for the Court, denied that corporations “as such” were persons as that term was used in Art. Ill and the Judiciary Act of 1789. See Bank of the United States v. Deveaux, 5 Cranch 61, 86 (1809).48 By 1844, however, the Deveaux doctrine was unhesitatingly abandoned:

“[A] corporation created by and doing business in a par*688ticular state, is to be deemed to all intents and purposes as a person, although an artificial person, . . . capable of being treated as a citizen of that state, as much as a natural person.” Louisville R. Co. v. Letson, 2 How. 497, 558 (1844) (emphasis added), discussed in Globe 752.

And only two years before the debates on the Civil Rights Act, in Cowles v. Mercer County, 7 Wall. 118, 121 (1869), the Letson principle was automatically and without discussion extended to municipal corporations. Under this doctrine, municipal corporations were routinely sued in the federal courts49 and this fact was well known to Members of Congress.50

That the “usual” meaning of the word “person” would extend to municipal corporations is also evidenced by an Act of Congress which had been passed only months before the Civil Rights Act was passed. This Act provided that

“in all acts hereafter passed . . . the word 'person’ may extend and be applied to bodies politic and corporate . . . unless the context shows that such words were intended to be used in a more limited sense.” Act of Feb. 25, 1871, § 2, 16 Stat. 431.

Municipal corporations in 1871 were included within the phrase “bodies politic and corporate” 51 and, accordingly, the *689“plain meaning” of § 1 is that local government bodies were to be included within the ambit of the persons who could be sued under § 1 of the Civil Rights Act. Indeed, a Circuit Judge, writing in 1873 in what is apparently the first reported case under § 1, read the Dictionary Act in precisely this way in a case involving a corporate plaintiff and a municipal defendant.52 See Northwestern Fertilizing Co. v. Hyde Park, 18 F. Cas. 393, 394 (No. 10,336) (CC ND Ill. 1873).53

*690I — I

Our analysis of the legislative history of the Civil Rights Act of 1871 compels the conclusion that Congress did intend municipalities and other local government units to be included ' among those persons to whom § 1983 applies.54 Local govern- ■ ing bodies,55 therefore, can be sued directly under § 1983 for monetary, declaratory, or injunctive relief where, as here, the action that is alleged to be unconstitutional implements or executes a policy statement, ordinance, regulation, or decision ^.officially adopted and promulgated by that body’s officers. \ Moreover, although the touchstone of the § 1983 action against a government body is an allegation that official policy is responsible for a deprivation of rights protected by the Constitution, local governments, like every other § 1983 “person,” by the very terms of the statute, may be sued for constitu*691tional deprivations visited pursuant to governmental “custom” even though such a custom has not received formal approval through the body’s official decisionmaking channels. As Mr. Justice Harlan, writing for the Court, said in Adickes v. S. H. Kress & Co., 398 U. S. 144, 167-168 (1970): “Congress included customs and usages [in § 1983] because of the persistent and widespread discriminatory practices of state officials .... Although not authorized by written law, such practices of state officials could well be so permanent and well settled as to constitute a 'custom or usage’ with the force of law.” 56

On the other hand, the language of § 1983, read against the background of the same legislative history, compels the conclusion that Congress did not intend municipalities to be held liable unless action pursuant to official municipal policy-/of some nature caused a constitutional tort. In particular, we conclude that a municipality cannot be held liable solely because it employs a tortfeasor — or, in other words, a munici- ^ pality cannot be held liable under § 1983 on a respondeat '■ superior theory.

We begin with the language of § 1983 as originally passed:

“[A]ny person who, under color of any law, statute, ordinance, regulation, custom, or usage of any State, shall subject, or cause to be subjected, any person ... to the deprivation of any rights, privileges, or immunities secured by the Constitution of the United States, shall, any such *692law, statute, ordinance, regulation, custom, or usage of the State to the contrary notwithstanding, be liable to the party injured in any action at law, suit in equity, or other proper proceeding for redress . . . 17 Stat. 13 (emphasis added).

The italicized language plainly imposes liability on a government that, under color of some official policy, “causes” an employee to violate another’s constitutional rights. At the same time, that language cannot be easily read to impose liability vicariously on governing bodies solely on the basis of the existence of an employer-employee relationship with a tortfeasor. Indeed, the fact that Congress did specifically provide that A’s tort became B’s liability if B “caused” A to subject another to a tort suggests that Congress did not intend § 1983 liability to attach where such causation was absent.57 See Rizzo v. Goode, 423 U. S. 362, 370-371 (1976).

*693Equally important, creation of a federal law of respondeat superior would have raised all the constitutional problems associated with the obligation to keep the peace, an obligation Congress chose not to impose because it thought imposition of such an obligation unconstitutional. To this day, there is disagreement about the basis for imposing liability on an employer for the torts of an employee when the sole nexus between the employer and the tort is the fact of the employer-employee relationship. See W. Prosser, Law of Torts § 69, p. 469 (4th ed. 1971). Nonetheless, two justifications tend to stand out. First is the common-sense notion that no matter how blameless an employer appears to be in an individual case, accidents might nonetheless be reduced if employers had to bear the cost of accidents. See, e. g., ibid.; 2 F. Harper & F. James, Law of Torts, §26.3, pp. 1368-1369 (1956). Second is the argument that the cost of accidents should be *694spread to the community as a whole on an insurance theory. See, e. g., id., § 26.5; Prosser, supra, at 459.58

The first justification is of the same sort that was offered for statutes like the Sherman amendment: “The obligation to make compensation for injury resulting from riot is, by arbitrary enactment of statutes, affirmatory law, and the reason of passing the statute is to secure a more perfect police regulation.” Globe 777 (Sen. Frelinghuysen). This justification was obviously insufficient to sustain the amendment against perceived constitutional difficulties and there is no reason to suppose that a more general liability imposed for a similar reason would have been thought less constitutionally objectionable. The second justification was similarly put forward as a justification for the Sherman amendment: “we do not look upon [the Sherman amendment] as a punishment .... It is a mutual insurance.” Id., at 792 (Rep. Butler). Again, this justification was insufficient to sustain the amendment.

We conclude, therefore, that a local government may not be sued under § 1983 for an injury inflicted solely by its employees or agents. Instead, it is when execution of a government’s policy or custom, whether made by its lawmakers or by those whose edicts or acts may fairly be said to represent official policy, inflicts the injury that the. government as an entity is responsible under § 1983. Since this case unquestionably involves official policy as the moving force of the constitutional violation found by the District Court, see supra, at *695660-662, and n. 2, we must reverse the judgment below. In so doing, we have no occasion to address, and do not address, what the full contours of municipal liability under § 1983 may be. We have attempted only to sketch so much of the § 1983 cause of action against a local government as is apparent from the history of the 1871 Act and our prior cases, and we expressly leave further development of this action to another day.

Ill

Although we have stated that stare decisis has more force in statutory analysis than in constitutional adjudication because, in the former situation, Congress can correct our mistakes through legislation, see, e. g., Edelman v. Jordan, 415 U. S. 651, 671, and n. 14 (1974), we have never applied stare decisis mechanically to prohibit overruling our earlier decisions determining the meaning of statutes. See, e. g., Continental T. V., Inc. v. GTE Sylvania Inc., 433 U. S. 36, 47-49 (1977); Burnet v. Coronado Oil & Gas Co., 285 U. S. 393, 406 n. 1 (1932) (Brandéis, J., dissenting) (collecting cases). Nor is this a case where we should “place on the shoulders of Congress the burden of the Court’s own error.” Girouard v. United States, 328 U. S. 61, 70 (1946).

First, Monroe v. Pape, insofar as it completely immunizes municipalities from suit under § 1983, was a departure from prior practice. See, e. g., Northwestern Fertilizing Co. v. Hyde Park, 18 F. Cas. 393 (No. 10,336) (CC ND Ill. 1873); City of Manchester v. Leiby, 117 F. 2d 661 (CA1 1941); Hannan v. City of Haverhill, 120 F. 2d 87 (CA1 1941); Douglas v. City of Jeannette, 319 U. S. 157 (1943); Holmes v. Atlanta, 350 U. S. 879 (1955), in each of which municipalities were defendants in § 1983 suits.59 Moreover, the constitutional de-*696feet that led to the rejection of the Sherman amendment would not have distinguished between municipalities and school boards, each of which is an instrumentality of state administration. See supra, at 673-682. For this reason, our cases — decided both before and after Monroe, see n. 5, supra— holding school boards liable in § 1983 actions are inconsistent with Monroe, especially as Monroe’s, immunizing principle was extended to suits for injunctive relief in City of Kenosha v. Bruno, 412 U. S. 507 (1973).60 And although in many of these cases jurisdiction was not questioned, we ought not “disregard the implications of an exercise of judicial authority assumed to be proper for [100] years.” Brown Shoe Co. v. United States, 370 U. S. 294, 307 (1962); see Bank of the United States v. Deveaux, 5 Cranch, at 88 (Marshall, C. J.) (“Those decisions are not cited as authority . . . but they have much weight, as they show that this point neither occurred to the bar or the bench”). Thus, while we have reaffirmed Monroe without further examination on three occasions,61 it can scarcely be said that Monroe is so consistent with the warp and woof of civil rights law as to be beyond question.

Second, the principle of blanket immunity established in Monroe cannot be cabined short of school boards. Yet such an extension would itself be inconsistent with recent expressions of congressional intent. In the wake of our decisions, Congress not only has shown no hostility to federal-court decisions against school boards, but it has indeed rejected efforts to strip the federal courts of jurisdiction over school boards.62 Moreover, recognizing that school boards are often *697defendants in school desegregation suits, which have almost without exception been § 1983 suits, Congress has twice passed legislation authorizing grants to school boards to assist them in complying with federal-court decrees.63 Finally, in *698regard to the Civil Rights Attorney’s Fees Awards Act of 1976, 90 Stat. 2641, 42 U. S. C. § 1988 (1976 ed.), which allows prevailing parties (in the discretion of the court) in § 1983 suits *699to obtain attorney’s fees from the losing parties, the Senate stated:

“[Defendants in these cases are often State or local bodies or State or local officials. In such cases it is intended that the attorneys’ fees, like other items of costs, will be collected either directly from the official, in his official capacity, from funds of his agency or under his control, or from the State or local government (whether or not the agency or government is a named party).” S. Rep. No. 94-1011, p. 5 (1976) (emphasis added; footnotes omitted).

Far from showing that Congress has relied on Monroe, therefore, events since 1961 show that Congress has refused to extend the benefits of Monroe to school boards and has attempted to allow awards of attorney’s fees against local governments even though Monroe, City of Kenosha v. Bruno, and Aldinger v. Howard, 427 U. S. 1 (1976), have made the joinder of such governments impossible.64

Third, municipalities can assert no reliance claim which can *700support an absolute immunity. As Mr. Justice Frankfurter said in Monroe, “[t]his is not an area of commercial law in which, presumably, individuals may have arranged their affairs in reliance on the expected stability of decision.” 365 U. S., at 221-222 (dissenting in part). Indeed, municipalities simply cannot “arrange their affairs” on an assumption that they can violate constitutional rights indefinitely since injunctive suits against local officials under § 1983 would prohibit any such arrangement. And it scarcely need be mentioned that nothing in Monroe encourages municipalities to violate constitutional rights or even suggests that such violations are anything other than completely wrong.

Finally, even under the most stringent test for the propriety of overruling a statutory decision proposed by Mr. Justice Harlan in Monroe65 — “that it appear beyond doubt from the legislative history of the 1871 statute that [Monroe] misapprehended the meaning of the [section],” 365 U. S., at 192 (concurring opinion) — the overruling of Monroe insofar as it holds that local governments are not “persons” who may be defendants in § 1983 suits is clearly proper. It is simply beyond doubt that, under the 1871 Congress’ view of the law, were § 1983 liability unconstitutional as to local governments, it would have been equally unconstitutional as to state officers. Yet everyone — proponents and opponents alike — knew § 1983 would be applied to state officers and nonetheless stated that § 1983 was constitutional. See supra, at 680-682. And, moreover, there can be no doubt that § 1 of the Civil Rights Act was intended to provide a remedy, to be broadly construed, against all forms of official violation of federally protected *701rights. Therefore, absent a clear statement in the legislative history supporting the conclusion that § 1 was not to apply to the official acts of a municipal corporation — which simply is not present — there is no justification for excluding municipalities from the “persons” covered by § 1.

For the reasons stated above, therefore, we hold that stare decisis does not bar our overruling of Monroe insofar as it is inconsistent with Parts I and II of this opinion.66

IV

Since the question whether local government bodies should be afforded some form of official immunity was not presented as a question to be decided on this petition and was not briefed by the parties or addressed by the courts below, we express no views on the scope of any municipal immunity beyond holding that municipal bodies sued under § 1983 cannot be entitled to an absolute immunity, lest our decision that such bodies are subject to suit under § 1983 “be drained of meaning,” Scheuer v. Rhodes, 416 U. S. 232, 248 (1974). Cf. Bivens v. Six Unknown Fed. Narcotics Agents, 403 U. S. 388, 397-398 (1971).

*702y

For the reasons stated above, the judgment of the Court of Appeals is

Reversed.

APPENDIX TO OPINION OF THE COURT

As proposed, the Sherman amendment was as follows:

“That if any house, tenement, cabin, shop, building, barn, or granary shall be unlawfully or feloniously demolished, pulled down, burned, or destroyed, wholly or in part, by any persons riotously and tumultuously assembled together; or if any person shall unlawfully and with force and violence be whipped, scourged, wounded, or killed by any persons riotously and tumultuously assembled together; and if such offense was committed to deprive any person of any right conferred upon him by the Constitution and laws of the United States, or to deter him or punish him for exercising such right, or by reason of his race, color, or previous condition of servitude, in every such case the inhabitants of the county, city, or parish in which any of the said offenses shall be committed shall be liable to pay full compensation to the person or persons damnified by such offense if living, or to his widow or legal representative if dead; and such compensation may be recovered by such person or his representative by a suit in any court of the United States of competent jurisdiction in the district in which the offense was committed, to be in the name of the person injured, or his legal representative, and against said county, city, or parish. And execution may be issued on a judgment rendered in such suit and may be levied upon any property, real or personal, of any person in said county, city, or parish, and the said county, city, or parish may recover the full amount of such judgment, costs and interest, *703from any person or persons engaged as principal or accessory in such, riot in an action in any court of competent jurisdiction.” Globe 663.

The complete text of the first conference substitute for the Sherman amendment is:

“That if any house, tenement, cabin, shop, building, barn, or granary shall be unlawfully or feloniously demolished, pulled down, burned, or destroyed, wholly or in part, by any persons riotously and tumultuously assembled together; or if any person shall unlawfully and with force and violence be whipped, scourged, wounded, or killed by any persons riotously and tumultuously assembled together, with intent to deprive any person of any right conferred upon him by the Constitution and laws of the United States, or to deter him or punish him for exercising such right, or by reason of his race, color, or previous condition of servitude, in every such case the county, city, or parish in which any of the said offenses shall be committed shall be liable to pay full compensation to the person or persons damnified by such offense, if living, or to his widow or legal representative if dead; and such compensation may be recovered in an action on the case by such person or his representative in any court of the United States of competent jurisdiction in the district in which the offense was committed, such action to be in the name of the person injured, or his legal representative, and against said county, city, or parish, and in which action any of the parties committing such acts may be joined as defendants. And any payment of any judgment, or part thereof unsatisfied, recovered by the plaintiff in such action, may, if not satisfied by the individual defendant therein within two months next after the recovery of such judgment upon execution duly issued against such individual defendant in such judgment, and returned unsatisfied, in whole or in part, be enforced *704against such county, city, or parish, by execution, attachment, mandamus, garnishment, or any other proceeding in aid of execution or applicable to the enforcement of judgments against municipal corporations; and such judgment shall be a lien as well upon all moneys in the treasury of such county, city, or parish, as upon the other property thereof. And the court in any such action may on motion cause additional parties to be made therein prior to issue joined, to the end that justice may be done. And the said county, city, or parish may recover the full amount of such judgment, by it paid, with costs and interest, from any person or persons engaged as principal or accessory in such riot, in an action in any court of competent jurisdiction. And such county, city, or parish, so paying, shall also be subrogated to all the plaintiff's rights under such judgment.” Id., at 749, 755.

The relevant text of the second conference substitute for the Sherman amendment is as follows:

“[A]ny person or persons having knowledge that any of the wrongs conspired to be done and mentioned in the second section of this act are about to be committed, and having power to prevent or aid in preventing the same, shall neglect or refuse so to do, and such wrongful act shall be committed, such person or persons shall be liable to the person injured, or his legal representatives.” Id., at 804 (emphasis added).

Me. Justice Powell,

concurring.

I join the opinion of the Court, and express these additional views.

New cases in the history of the Court have been cited more frequently than Monroe v. Pape, 365 U. S. 167 (1961), decided less than two decades ago. Focusing new light on 42 U. S. C. § 1983, that decision widened access to the federal courts and permitted expansive interpretations of the reach of *705the 1871 measure. But Monroe exempted local governments from liability at the same time it opened wide the courthouse door to suits against officers and employees of those entities— even when they act pursuant to express authorization. The oddness of this result, and the weakness of the historical evidence relied on by the Monroe Court in support of it, are well demonstrated by the Court’s opinion today. Yet the gravity of overruling a part of so important a decision prompts me to write.

I

In addressing a complaint alleging unconstitutional police conduct that probably was unauthorized and actionable under state law,1 the Monroe Court treated the 42d Congress’ rejection of the Sherman amendment as conclusive evidence of an intention to immunize local governments from all liability under the statute for constitutional injury. That reading, in light of today’s thorough canvass of the legislative history, clearly “misapprehended the meaning of the controlling provision,” Monroe, supra, at 192 (Harlan, J., concurring). In this case, involving formal, written policies of the Department of Social Services and the Board of Education of the city of New York that are alleged to conflict *706with the command of the Due Process Clause, cf. Cleveland Board of Education v. LaFleur, 414 U. S. 632 (1974), the Court decides “not to reject [wisdom] merely because it comes late,” Henslee v. Union Planters Bank, 335 U. S. 595, 600 (1949) (Frankfurter, J., dissenting).

As the Court demonstrates, the Sherman amendment presented an extreme example of “riot act” legislation that sought to impose vicarious liability on government subdivisions for the consequences of private lawlessness. As such, it implicated concerns that are of marginal pertinence to the operative principle of § 1 of the 1871 legislation — now § 1983 — that “any person” acting “under color of” state law may be held liable for affirmative conduct that “subjects, or causes to be subjected, any person ... to the deprivation of any” federal constitutional or statutory right. Of the many reasons for the defeat of the Sherman proposal, none supports Monroe’s observation that the 42d Congress was fundamentally “antagonistic,” 365 U. S., at 191, to the proposition that government entities and natural persons alike should be held accountable for the consequences of conduct directly working a constitutional violation. Opponents in the Senate appear to have been troubled primarily by the proposal’s unprecedented lien provision, which would have exposed even property held for public purposes to the demands of § 1983 judgment lienors. Ante, at 673-674, n. 30. The opposition in the House of Representatives focused largely on the Sherman amendment’s attempt to impose a peacekeeping obligation on municipalities when the Constitution itself imposed no such affirmative duty and when many municipalities were not even empowered under state law to maintain police forces. Ante, at 673-675, 679-682.2

*707The Court correctly rejects a view of the legislative history that would produce the anomalous result of immunizing local government units from monetary liability for action directly causing a constitutional deprivation, even though such actions may be fully consistent with, and thus not remediable under, state law. No conduct of government comes more clearly within the “under color of” state law language of § 1983. It is most unlikely that Congress intended public officials acting under the command or the specific authorization of the government employer to be exclusively liable for resulting constitutional injury.3

As elaborated in Part II of today’s opinion, the rejection of the Sherman amendment can best be understood not as evidence of Congress’ acceptance of a rule of absolute municipal immunity but as a limitation of the statutory ambit to actual wrongdoers, i. e., a rejection of respondeat superior or any other principle of vicarious liability. Cf. Levin, The Section 1983 Municipal Immunity Doctrine, 65 Geo. L. J. 1483, 1531-1535 (1977). Thus, it has been clear that a public official may be held liable in damages when his actions are found to violate a constitutional right and there is no qualified immunity, see Wood v. Strickland, 420 U. S. 308 (1975); Procunier v. Navarette, 434 U. S. 555 (1978). Today the Court recognizes *708that this principle also applies to a local government when implementation of its official policies or established customs inflicts the constitutional injury.

II

This Court traditionally has been hesitant to overrule prior constructions of statutes or interpretations of common-law rules. “Stare decisis is usually the wise policy,” Burnet v. Coronado Oil & Gas Co., 285 U. S. 393, 406 (1932) (Brandéis, J., dissenting), but this cautionary principle must give way to countervailing considerations in appropriate circumstances.4 I concur in the Court’s view that this is not a case where we should “place on the shoulders of Congress the burden of the Court’s own error.” Girouard v. United States, 328 U. S. 61, 70 (1946).

Nor is this the usual case in which the Court is asked to overrule a precedent. Here considerations of stare decisis cut in both directions. On the one hand, we have a series of rulings that municipalities and counties are not “persons” for purposes of § 1983. On the other hand, many decisions of this Court have been premised on the amenability of school boards and similar entities to § 1983 suits.

In Monroe and its progeny, we have answered a question that was never actually briefed or argued in this Court— whether a municipality is liable in damages for injuries that are the direct result of its official policies. “The theory of the complaint [in Monroe was] that under the circumstances [t]here alleged the City [was] liable for the acts of its police officers, by virtue of respondeat superior.” Brief for Petition*709ers, O. T. 1960, No. 39, p. 21.5 Respondents answered that adoption of petitioners’ position would expose “Chicago and every other municipality in the United States ... to Civil Rights Act liability through no action of its own and based on action contrary to its own ordinances and the laws of the state it is a part of.” Brief for Respondents, O. T. 1960, No. 39, p. 26. Thus the ground of decision in Monroe was not advanced by either party and was broader than necessary to resolve the contentions made in that case.6

*710Similarly, in Moor v. County of Alameda, 411 U. S. 693 (1973), petitioners asserted that “the County was vicariously liable for the acts of its deputies and sheriff,” id., at 696, under 42 U. S. C. § 1988. In rejecting this vicarious-liability claim, 411 U. S., at 710, and n. 27, we reaffirmed Monroe’s reading of the statute, but there was no challenge in that case to “the holding in Monroe concerning the status under § 1983 of public entities such as the County,” 411 U. S., at 700; Brief for Petitioners, O. T. 1972, No. 72-10, p. 9.

Only in City of Kenosha v. Bruno, 412 U. S. 607 (1973), did the Court confront a § 1983 claim based on conduct that was both authorized under state law and the direct cause of the claimed constitutional injury. In Kenosha, however, we raised the issue of the city’s amenability to suit under § 1983 on our own initiative.7

This line of cases — from Monroe to Kenosha — is difficult to reconcile on a principled basis with a parallel series of cases *711in which the Court has assumed sub silentio that some local government entities could be sued under § 1983. If now, after full consideration of the question, we continued to adhere to Monroe, grave doubt would be cast upon the Court’s exercise of § 1983 jurisdiction over school boards. See ante, at 663 n. 5. Since “the principle of blanket immunity established in Monroe cannot be cabined short of school boards,” ante, at 696, the conflict is squarely presented. Although there was an independent basis of jurisdiction in many of the school board cases because of the inclusion of individual public officials as nominal parties, the opinions of this Court make explicit reference to the school board party, particularly in discussions of the relief to be awarded, see, e. g., Green v. County School Board, 391 U. S. 430, 437-439, 441-442 (1968); Milliken v. Bradley, 433 U. S. 267, 292-293 (1977) (Powell, J., concurring in judgment). And, as the Court points out, ante, at 696-697, and nn. 62, 63, Congress has focused specifically on this Court’s school board decisions in several statutes. Thus the exercise of § 1983 jurisdiction over school boards, while perhaps not premised on considered holdings, has been longstanding. Indeed, it predated Monroe.

Even if one attempts to explain away the school board decisions as involving suits which “may be maintained against board members in their official capacities for injunctive relief under either § 1983 or Ex parte Young, 209 U. S. 123 (1908),” post, at 716-717, n. 2, some difficulty remains in rationalizing the relevant body of precedents. At least two of the school board cases involved claims for monetary relief. Cohen v. Chesterfield County School Board, 326 F. Supp. 1159, 1161 (ED Va. 1971), rev’d, 474 F. 2d 395 (CA4 1973), rev’d and remanded, 414 U. S. 632 (1974); Tinker v. Des Moines Independent School Dist., 393 U. S. 503, 504 (1969). See also Vlandis v. Kline, 412 U. S. 441, 445 (1973). Although the point was not squarely presented in this Court, these claims *712for damages could not have been maintained in official-capacity suits if the government entity were not itself suable. Cf. Edelman v. Jordan, 415 U. S. 651 (1974).8 Moreover, the rationale of Kenosha would have to be disturbed to avoid closing all avenues under § 1983 to injunctive relief against constitutional violations by local government. The Court of Appeals in this case suggested that we import, by analogy, the Eleventh Amendment fiction of Ex parte Young into § 1983, 532 F. 2d 259, 264—266 (CA2 1976). That approach, however, would create tension with Kenosha because it would require “a bifurcated application” of “the generic word ‘person’ in § 1983” to public officials “depending on the nature of the relief sought against them.” 412 U. S., at 513. A public official sued in his official capacity for carrying out official policy would be a “person” for purposes of injunctive relief, but a non-“person” in an action for damages. The Court’s holding avoids this difficulty. See ante, at 690 n. 55.

Finally, if we continued to adhere to a rule of absolute municipal immunity under § 1983, we could not long avoid the question whether “we should, by analogy to our decision in Bivens v. Six Unknown Fed. Narcotics Agents, 403 U. S. 388 (1971), imply a cause of action directly from the Fourteenth Amendment which would not be subject to the limitations contained in § 1983 . . . .” Mt. Healthy City Board of Ed. v. Doyle, 429 U. S. 274, 278 (1977). One aspect of that inquiry would be whether there are any “special factors counselling hesitation in the absence of affirmative action by Congress,” Bivens v. Six Unknown Fed. Narcotics Agents, 403 U. S. 388, 396 (1971), such as an “explicit congressional declaration *713that persons injured by a [municipality] may not recover money damages . . . , but must instead be remitted to another remedy, equally effective in the view of Congress,” id., at 397. In light of the Court’s persuasive re-examination in today’s decision of the 1871 debates, I would have difficulty inferring from § 1983 “an explicit congressional declaration” against municipal liability for the implementation of official policies in violation of the Constitution. Rather than consti-tutionalize a cause of action against local government that Congress intended to create in 1871, the better course is to confess error and set the record straight, as the Court does today.9

Ill

Difficult questions nevertheless remain for another day. There are substantial line-drawing problems in determining “when execution of a government’s policy or custom” can be said to inflict constitutional injury such that “government as an entity is responsible under § 1983.” Ante, at 694. This case, however, involves formal, written policies of a municipal department and school board; it is the clear case. The Court also reserves decision on the availability of a qualified municipal immunity. Ante, at 701. Initial resolution of the question whether the protection available at common law for municipal corporations, see post, at 720-721, or other principles support a *714qualified municipal immunity in the context of the § 1983 damages action, is left to the lower federal courts.

Mr. Justice Stevens,

concurring in part.

Since Parts II and IV of the opinion of the Court are merely advisory and are not necessary to explain the Court’s decision, I join only Parts I, III, and V.

Mr. Justice Rehnquist,

dissenting.

Seventeen years ago, in Monroe v. Pape, 365 U. S. 167 (1961), this Court held that the 42d Congress did not intend to subject a municipal corporation to liability as a “person” within the meaning of 42 U. S. C. § 1983. Since then, the Congress has remained silent, but this Court has reaffirmed that holding on at least three separate occasions. Aldinger v. Howard, 427 U. S. 1 (1976); City of Kenosha v. Bruno, 412 U. S. 507 (1973); Moor v. County of Alameda, 411 U. S. 693 (1973). See also Mt. Healthy City Board of Ed. v. Doyle, 429 U. S. 274, 277-279 (1977). Today, the Court abandons this long and consistent line of precedents, offering in justification only an elaborate canvass of the same legislative history which was before the Court in 1961. Because I cannot agree that this Court is “free to disregard these precedents,” which have been “considered maturely and recently” by this Court, Runyon v. McCrary, 427 U. S. 160, 186 (1976) (Powell, J., concurring), I am compelled to dissent.

I

As this Court has repeatedly recognized, id., at 175 n. 12; Edelman v. Jordan, 415 U. S. 651, 671 n. 14 (1974), considerations of stare decisis are at their strongest when this Court confronts its previous constructions of legislation. In all cases, private parties shape their conduct according to this Court’s settled construction of the law, but the Congress is at *715liberty to correct our mistakes of statutory construction, unlike our constitutional interpretations, whenever it sees fit. The controlling principles were best stated by Mr. Justice Brandéis:

“Stare decisis is usually the wise policy, because in most matters it is more important that the applicable rule of law be settled than that it be settled right. . . . This is commonly true even where the error is a matter of serious concern, provided correction can be had by legislation. But in cases involving the Federal Constitution, where correction through legislative action is practically impossible, this Court has often overruled its earlier decisions.” Burnet v. Coronado Oil & Gas Co., 285 U. S. 393, 406-407 (1932) (dissenting opinion) (footnotes omitted).

Only the most compelling circumstances can justify this Court’s abandonment of such firmly established statutory precedents. The best exposition of the proper burden of persuasion was delivered by Mr. Justice Harlan in Monroe itself:

“From my point of view, the policy of stare decisis, as it should be applied in matters of statutory construction, and, to a lesser extent, the indications of congressional acceptance of this Court’s earlier interpretation, require that it appear beyond doubt from the legislative history of the 1871 statute that [United States v.] Classic, [313 U. S. 299 (1941)] and Screws [v. United States, 325 U. S. 91 (1945)] misapprehended the meaning of the controlling provision, before a departure from what was decided in those cases would be justified.” 365 U. S., at 192 (concurring opinion) (footnote omitted; emphasis added).

The Court does not demonstrate that any exception to this general rule is properly applicable here. The Court’s first assertion, that Monroe “was a departure from prior practice,” ante, at 695, is patently erroneous. Neither in Douglas v. City of Jeannette, 319 U. S. 157 (1943), nor in Holmes v. Atlanta, *716350 U. S. 879 (1955), nor in any of the school board cases cited by the Court, ante, at 663 n. 5, was the question now before us raised by any of th'e litigants or addressed by this Court. As recently as four Terms ago, we said in Hagans v. Lavine, 415 U. S. 528, 535 n. 5 (1974):

“Moreover, when questions of jurisdiction have been passed on in prior decisions sub silentio, this Court has never considered itself bound when a subsequent case finally brings the jurisdictional issue before us.”

The source of this doctrine that jurisdictional issues decided sub silentio are not binding in other cases seems to be Mr. Chief Justice Marshall's remark in United States v. More, 3 Cranch 159, 172 (1805).1 While the Chief Justice also said that such decisions may “have much weight, as they show that this point neither occurred to the bar or the bench,” Bank of the United States v. Deveaux, 5 Cranch 61, 88 (1809), unconsidered assumptions of jurisdiction simply cannot outweigh four consistent decisions of this Court, explicitly considering and rejecting that jurisdiction.

Nor is there any indication that any later Congress has ever approved suit against any municipal corporation under § 1983. Of all its recent enactments, only the Civil Rights Attorney’s Fees Awards Act of 1976, § 2, 90 Stat. 2641, 42 U. S. C. § 1988 (1976 ed.), explicitly deals with the Civil Rights Act of 1871.2 The 1976 Act provides that attorney’s fees may be awarded *717to the prevailing party “[i]n any action or proceeding to enforce a provision of sections 1981, 1982, 1983, 1985, and 1986 of this title.” There is plainly no language in the 1976 Act which would enlarge the parties suable under those substantive sections; it simply provides that parties who are already suable may be made liable for attorney’s fees. As the Court admits, ante, at 699, the language in the Senate Report stating that liability may be imposed “whether or not the agency or government is a named party,” S. Rep. No. 94-1011, p. 5 (1976), suggests that Congress did not view its purpose as being in any way inconsistent with the well-known holding of Monroe.

The Court’s assertion that municipalities have no right to act “on an assumption that they can violate constitutional rights indefinitely,” ante, at 700, is simply beside the point. Since Monroe, municipalities have had the right to expect that they would not be held liable retroactively for their officers’ failure to predict this Court’s recognition of new constitutional rights. No doubt innumerable municipal insurance policies and indemnity ordinances have been founded on this assumption, which is wholly justifiable under established principles of stare decisis. To obliterate those legitimate expectations without more compelling justifications than those advanced by the Court is a significant departure from our prior practice.

I cannot agree with Mr. Justice Powell’s view that “[w]e owe somewhat less deference to a decision that was rendered without benefit of a full airing of all the relevant considerations.” Ante, at 709 n. 6. Private parties must be able to rely upon explicitly stated holdings of this Court without being *718obliged to peruse the briefs of the litigants to predict the likelihood that this Court might change its mind. To cast such doubt upon each of our cases, from Marbury v. Madison, 1 Cranch 137 (1803), forward, in-which the explicit ground of decision "was never actually briefed or argued,” ante, at 708 (Powell, J., concurring), would introduce intolerable uncertainty into the law. Indeed, in Marbury itself, the argument of Charles Lee on behalf of the applicants — which, unlike the arguments in Monroe, is reproduced in the Reports of this Court where anyone can see it — devotes not a word to the question of whether this Court has the power to invalidate a statute duly enacted by the Congress. Neither this ground of decision nor any other was advanced by Secretary of State Madison, who evidently made no appearance. 1 Cranch, at 153-154. More recent landmark decisions of this Court would appear to be likewise vulnerable under my Brother Powell’s analysis. In Mapp v. Ohio, 367 U. S. 643 (1961), none of the parties requested the Court to overrule Wolf v. Colorado, 338 U. S. 25 (1949); it did so only at the request of an amicus curiae. 367 U. S., at 646 n. 3. That Marbury, Mapp, and countless other decisions retain their vitality despite their obvious flaws is a necessary byproduct of the adversary system, in which both judges and the general public rely upon litigants to present “all the relevant considerations.” Ante, at 709 n. 6 (Powell, J., concurring). While it undoubtedly has more latitude in the field of constitutional interpretation, this Court is surely not free to abandon settled statutory interpretation at any time a new thought seems appealing.3

Thus, our only task is to discern the intent of the 42d Congress. That intent was first expounded in Monroe, and it *719has been followed consistently ever since. This is not some esoteric branch of the law in which congressional silence might reasonably be equated with congressional indifference. Indeed, this very year, the Senate has been holding hearings on a bill, S. 35, 95th Cong., 1st Sess. (1977), which would remove the municipal immunity recognized by Monroe. 124 Cong. Rec. D117 (daily ed. Feb. 8, 1978). In these circumstances, it cannot be disputed that established principles of stare decisis require this Court to pay the highest degree of deference to its prior holdings. Monroe may not be overruled unless it has been demonstrated “beyond doubt from the legislative history of the 1871 statute that [Monroe] misapprehended the meaning of the controlling provision.” Monroe, 365 U. S., at 192 (Harlan, J., concurring). The Court must show not only that Congress, in rejecting the Sherman amendment, concluded that municipal liability was not unconstitutional, but also that, in enacting § 1, it intended to impose that liability. I am satisfied that no such showing has been made.

II

Any analysis of the meaning of the word “person” in § 1983, which was originally enacted as § 1 of the Ku Klux Klan Act of April 20, 1871, 17 Stat. 13, must begin, not with the Sherman amendment, but with the Dictionary Act. The latter Act, which supplied rules of construction for all legislation, provided:

“That in all acts hereafter passed . . . the word ‘person’ may extend and be applied to bodies politic and corporate . . . unless the context shows that such words were intended to be used in a more limited sense . . . .” Act of Feb. 25, 1871, § 2, 16 Stat. 431.

The Act expressly provided that corporations need not be included within the scope of the word “person” where the coñ-text suggests a more limited reach. Not a word in the legislative history of the Act gives any indication of the contexts *720in which Congress felt it appropriate to include a corporation as a person. Indeed, the chief cause of concern was that the Act’s provision that “words importing the masculine gender may be applied to females,” might lead to an inadvertent extension of the suffrage to women. Cong. Globe, 41st Cong., 3d Sess., 777 (1871) (remarks of Sen. Sawyer).

There are other factors, however, which suggest that the Congress which enacted § 1983 may well have intended the word “person” “to be used in a more limited sense,” as Monroe concluded. It is true that this Court had held that both commercial corporations, Louisville R. Co. v. Letson, 2 How. 497, 558 (1844), and municipal corporations, Cowles v. Mercer County, 7 Wall. 118, 121 (1869), were “citizens” of a State within the meaning of the jurisdictional provisions of Art. III. Congress, however, also knew that this label did not apply in all contexts, since this Court, in Paul v. Virginia, 8 Wall. 168 (1869), had held commercial corporations not to be “citizens” within the meaning of the Privileges and Immunities Clause, U. S. Const., Art. IV, § 2. Thus, the Congress surely knew that, for constitutional purposes, corporations generally enjoyed a different status in different contexts. Indeed, it may be presumed that Congress intended that a corporation should enjoy the same status under the Ku Klux Klan Act as it did under the Fourteenth Amendment, since it had been assured that § 1 “was so very simple and really reenact [ed] the Constitution.” Cong. Globe, 42d Cong., 1st Sess., 569 (1871) (remarks of Sen. Edmunds). At the time § 1983 was enacted the only federal case to consider the status of corporations under the Fourteenth Amendment had concluded, with impeccable logic, that a corporation was neither a “citizen” nor a “person.” Insurance Co. v. New Orleans, 13 F. Cas. 67 (No. 7,052) (CC La. 1870).

Furthermore, the state courts did not speak with a single voice with regard to the tort liability of municipal corporations. Although many Members of Congress represented *721States which had retained absolute municipal tort immunity, see, e. g., Irvine v. Town of Greenwood, 89 S. C. 511, 72 S. E. 228 (1911) (collecting earlier cases), other States had adopted the currently predominant distinction imposing liability for proprietary acts, see generally 2 F. Harper & F. James, Law of Torts § 29.6 (1956), as early as 1842, Bailey v. Mayor of City of New York, 3 Hill 531 (N. Y. 1842). Nevertheless, no state court had ever held that municipal corporations were always liable in tort in precisely the same manner as other persons.

The general remarks from the floor on the liberal purposes of § 1 offer no explicit guidance as to the parties against whom the remedy could be enforced. As the Court concedes, only Representative Bingham raised a concern which could be satisfied only by relief against governmental bodies. Yet he never directly related this concern to § 1 of the Act. Indeed, Bingham stated at the outset, "I do not propose now to discuss the provisions of the bill in detail,” Cong. Globe, 42d Cong., 1st Sess., App. 82 (1871), and, true to his word, he launched into an extended discourse on the beneficent purposes of the Fourteenth Amendment. While Bingham clearly stated that Congress could “provide that no citizen in any State shall be deprived of his property by State law or the judgment of a State court without just compensation therefor,” id., at 85, he never suggested that such a power was exercised in § l.4 *722Finally, while Bingham has often been advanced as the chief expositor of the Fourteenth Amendment, Duncan v. Louisiana, 391 U. S. 145, 165 (1968) (Black, J., concurring); Adamson v. California, 332 U. S. 46, 73-74 (1947) (Black, J., dissenting), there is nothing to indicate that his colleagues placed any greater credence in his theories than has this Court. See Duncan, supra, at 174-176 (Harlan, J., dissenting); Adamson, supra, at 64 (Frankfurter, J., concurring).

Thus, it ought not lightly to be presumed, as the Court does today, ante, at 690 n. 53, that § 1983 “should prima facie be construed to include ‘bodies politic’ among the entities that could be sued.” Neither the Dictionary Act, the ambivalent state of judicial decisions, nor the floor debate on § 1 of the Act gives any indication that any Member of Congress had any inkling that § 1 could be used to impose liability on municipalities. Although Senator Thurman, as the Court emphasizes, ante, at 686 n. 45, expressed his belief that the terms of § 1 “are as comprehensive as can be used,” Cong. Globe, 42d Cong., 1st Sess., App. 217 (1871), an examination of his lengthy remarks demonstrates that it never occurred to him that § 1 did impose or could have imposed any liability upon municipal corporations. In an extended parade of hor-ribles, this “old Roman,” who was one of the Act’s most implacable opponents, suggested that state legislatures, Members of Congress, and state judges might be held liable under the Act. Ibid. If, at that point in the debate, he had any idea that § 1 was designed to impose tort liability upon cities and counties, he would surely have raised an additional outraged objection. Only once was that possibility placed squarely before the Congress — in its consideration of the Sherman amendment — and the Congress squarely rejected it.

The Court is probably correct that the rejection of the Sherman amendment does not lead ineluctably to the conclusion that Congress intended municipalities to be immune from liability under all circumstances. Nevertheless, it cannot be *723denied that the debate on that amendment, the only explicit consideration of municipal tort liability, sheds considerable light on the Congress’ understanding of the status of municipal corporations in that context. Opponents of the amendment were well aware that municipalities had been subjected to the jurisdiction of the federal courts in the context of suits to enforce their contracts, Cong. Globe, 42d Cong., 1st Sess., 789 (1871) (remarks of Rep. Kerr), but they expressed their skepticism that such jurisdiction should be exercised in cases sounding in tort:

“Suppose a judgment obtained under this section, and no property can be found to levy upon except the courthouse, can we levy on the court-house and sell it? So this section provides, and that too in an action of tort, in an action ex delicto, where the county has never entered into any contract, where the State has never authorized the county to assume any liability of the sort or imposed any liability upon it. It is in my opinion simply absurd.” Id., at 799 (remarks of Rep. Farnsworth).

Whatever the merits of the constitutional arguments raised against it, the fact remains that Congress rejected the concept of municipal tort liability on the only occasion in which the question was explicitly presented. Admittedly this fact is not conclusive as to whether Congress intended § 1 to embrace a municipal corporation within the meaning of “person,” and thus the reasoning of Monroe on this point is subject to challenge. The meaning of § 1 of the Act of 1871 has been subjected in this case to a more searching and careful analysis than it was in Monroe, and it may well be that on the basis of this closer analysis of the legislative debates a conclusion contrary to the Monroe holding could have been reached when that case was decided 17 years ago. But the rejection of the Sherman amendment remains instructive in that here alone did the legislative debates squarely focus on the liability of municipal corporations, and that liability was rejected. *724Any inference which might be drawn from the Dictionary Act or from general expressions of benevolence in the debate on § 1 that the word “person” was intended to include municipal corporations falls far short of showing “beyond doubt” that this Court in Monroe “misapprehended the meaning of the controlling provision.” Errors such as the Court may have fallen into in Monroe do not end the inquiry as to stare decisis; they merely begin it. I would adhere to the holding of Monroe as to the liability of a municipal corporation under § 1983.

Ill

The decision in Monroe v. Pape was the fountainhead of the torrent of civil rights litigation of the last 17 years. Using § 1983 as a vehicle, the courts have articulated new and previously unforeseeable interpretations of the Fourteenth Amendment. At the same time, the doctrine of municipal immunity enunciated in Monroe has protected municipalities and their limited treasuries from the consequences of their officials’ failure to predict the course of this Court’s constitutional jurisprudence. None of the Members of this Court can foresee the practical consequences of today’s removal of that protection. Only the Congress, which has the benefit of the advice of every segment of this diverse Nation, is equipped to consider the results of such a drastic change in the law. It seems all but inevitable that it will find it necessary to do so after today’s decision.

I would affirm the judgment of the Court of Appeals.

Mr. Justice Brennan

delivered the opinion of the Court.

Monell v. New York City Dept. of Social Services, 436 U. S. 658 (1978), overruled Monroe v. Pape, 365 U. S. 167 (1961), insofar as Monroe held that local governments were not among the “persons” to whom 42 U. S. C. 11983 applies and were therefore wholly immune from suit under the statute.1 Monell reserved decision, however, on the question whether local governments, although not entitled to an absolute immunity, should be afforded some form of official immunity in 1 1983 suits. 436 U. S., at 701. In this action brought by petitioner in the District Court for the Western District of Missouri, the Court of Appeals for the Eighth Circuit held that respondent city of Independence, Mo., “is entitled to qualified immunity from liability” based on the good faith *625of its officials: “We extend the limited immunity the district court applied to the individual defendants to cover the City as well, because its officials acted in good faith and without malice.” 589 F. 2d 335, 337-338 (1978). We granted certiorari. 444 U. S. 822 (1979). We reverse.

I

The events giving rise to this suit are detailed in the District Court’s findings of fact, 421 F. Supp. 1110 (1976). On February 20, 1967, Robert L. Broucek, then City Manager of respondent city of Independence, Mo,, appointed petitioner George D. Owen to an indefinite term as Chief of Police.2 In 1972, Owen and a new City Manager, Lyle W. Alberg, engaged in a dispute over petitioner’s administration of the Police Department’s property room. In March of that year, a handgun, which the records of the Department’s property room stated had been destroyed, turned up in Kansas City in the possession of a felon. This discovery prompted Al-berg to initiate an investigation of the management of the property room. Although the probe was initially directed by petitioner, Alberg soon transferred responsibility for the investigation to the city’s Department of Law, instructing the City Counselor to supervise its conduct and to inform him directly of its findings.

Sometime in early April 1972, Alberg received a written report on the investigation’s progress, along with copies of confidential witness statements. Although the City Auditor found that the Police Department’s records were insufficient to permit an adequate accounting of the goods contained in the property room, the City Counselor concluded that there was no evidence of any criminal acts or of any violation of *626state or municipal law in the administration of the property-room. Alberg discussed the results of the investigation at an informal meeting with several City Council members and advised them that he would take action at an appropriate time to correct any problems in the administration of the Police Department.

On April 10, Alberg asked petitioner to resign as Chief of Police and to accept another position within the Department, citing dissatisfaction with the manner in which petitioner had managed the Department, particularly his inadequate supervision of the property room. Alberg warned that if petitioner refused to take another position in the Department his employment would be terminated, to which petitioner responded that he did not intend to resign.

On April 13, Alberg issued a public statement addressed to the Mayor and the City Council concerning the results of the investigation. After referring to “discrepancies” found in the administration, handling, and security of public property, the release concluded that “[t]here appears to be no evidence to substantiate any allegations of a criminal nature” and offered assurances that “[sjteps have been initiated on an administrative level to correct these discrepancies.” Id., at 1115. Although Alberg apparently had decided by this time to replace petitioner as Police Chief, he took no formal action to that end and left for a brief vacation without informing the City Council of his decision.3

While Alberg was away on the weekend of April 15 and 16, two developments occurred. Petitioner, having consulted with counsel, sent Alberg a letter demanding written notice of the charges against him and a public hearing with a reason*627able opportunity to respond to those charges.4 At approximately the same time, City Councilman Paul L. Roberts asked for a copy of the investigative report on the Police Department property room. Although petitioner’s appeal received no immediate response, the Acting City Manager complied with Roberts’ request and supplied him with the audit report and witness statements.

On the evening of April 17, 1972, the City Council held its regularly scheduled meeting. After completion of the planned agenda, Councilman Roberts read a statement he had prepared on the investigation.5 Among other allegations, *628Roberts charged that petitioner had misappropriated Police Department property for his own use, that narcotics and money had “mysteriously disappeared” from his office, that traffic tickets had been manipulated, that high ranking police officials had made “inappropriate” requests affecting the police court, and that “things have occurred causing the unusual release of felons.” At the close of his statement, Roberts moved that the investigative reports be released to the news media and turned over to the prosecutor for presentation to the grand jury, and that the City Manager “take all direct *629and appropriate action” against those persons “involved in illegal, wrongful, or gross inefficient activities brought out in the investigative reports.” After some discussion, the City Council passed Roberts’ motion with no dissents and one abstention.6

City Manager Alberg discharged petitioner the very next day. Petitioner was not given any reason for his dismissal; he received only a written notice stating that his employment as Chief of Police was “[t]erminated under the provisions of Section 3.3(1) of the City Charter.”7 Petitioner’s earlier demand for a specification of charges and a public hearing was ignored, and a subsequent request by his attorney for an appeal of the discharge decision was denied by the city on the grounds that “there is no appellate procedure or forum provided by the Charter or ordinances of the City of Independence, Missouri, relating to the dismissal of Mr. Owen.” App. 26-27.

The local press gave prominent coverage both to the City Council’s action and petitioner’s dismissal, linking the discharge to the investigation.8 As instructed by the City Council, Alberg referred the investigative reports and witness statements to the Prosecuting Attorney of Jackson County, Mo., *630for consideration by a grand jury. The results of the audit and investigation were never released to the public, however. The grand jury subsequently returned a “no true bill,” and no further action was taken by either the City Council or City Manager Alberg.

II

Petitioner named the city of Independence, City Manager Alberg, and the present members of the City Council in their official capacities as defendants in this suit.9 Alleging that he was discharged without notice of reasons and without a hearing in violation of his constitutional rights to procedural and substantive due process, petitioner sought declaratory and injunctive relief, including a hearing on his discharge, back-pay from the date of discharge, and attorney’s fees. The District Court, after a bench trial, entered judgment for respondents. 421 F. Supp. 1110 (1976).10

*631The Court of Appeals initially reversed the District Court. 560 F. 2d 925 (1977).11 Although it agreed with the District Court that under Missouri law petitioner possessed no property interest in continued employment as Police Chief, the Court of Appeals concluded that the city’s allegedly false public accusations had blackened petitioner’s name and reputation, thus depriving him of liberty without due process of law. That the stigmatizing charges did not come from the City Manager and were not included in the official discharge notice was, in the court’s view, immaterial. What was un-*632portant, the court explained, was that "the official actions of the city council released charges against [petitioner] contemporaneous and, in the eyes of the public, connected with that discharge.” Id., at 937.12

Respondents petitioned for review of the Court of Appeals’ decision. Certiorari was granted, and the case was remanded for further consideration in light of our supervening decision in Monell v. New York City Dept. of Social Services, 436 U. S. 658 (1978). 438 U. S. 902 (1978). The Court of Ap*633peals on the remand reaffirmed its original determination that the city had violated petitioner’s rights under the Fourteenth Amendment, but held that all respondents, including the city, were entitled to qualified immunity from liability. 589 F. 2d 335 (1978).

Monell held that “a local government may not be sued under § 1983 for an injury inflicted solely by its employees or agents. Instead, it is when execution of a government’s policy or custom, whether made by its lawmakers or by those whose edicts or acts may fairly be said to represent official policy, inflicts the injury that the government as an entity is responsible under § 1983.” 436 U. S., at 694. The Court of Appeals held in the instant ease that the municipality’s official policy was responsible for the deprivation of petitioner’s constitutional rights: “[T]he stigma attached to [petitioner] in connection with his discharge was caused by the official conduct of the City’s lawmakers, or by those whose acts may fairly be said to represent official policy. Such conduct amounted to official policy causing the infringement of [petitioner’s] constitutional rights, in violation of section 1983.” 589 F. 2d, at 337.13

*634Nevertheless, the Court of Appeals affirmed the judgment of the District Court denying petitioner any relief against the respondent city, stating:

“The Supreme Court’s decisions in Board of Regents v. Roth, 408 U. S. 564 . . . (1972), and Perry v. Sindermann, 408 U. S. 593 . . . (1972), crystallized the rule establishing the right to a name-clearing hearing for a government employee allegedly stigmatized in the course of his discharge. The Court decided those two cases two months after the discharge in the instant case. Thus, officials of the City of Independence could not have been aware of [petitioner’s] right to a name-clearing , hearing in connection with the discharge. The City of Independence should not be charged with predicting the future course of constitutional law. We extend the limited immunity the district court applied to the individual defendants to cover the City as well, because its officials acted in good faith and without malice. We hold the City not liable for actions it could not reasonably have known violated [petitioner’s] constitutional rights.” Id., at 338 (footnote and citations omitted).14

*635We turn now to the reasons for our disagreement with this holding.15

Ill

Because the question of the scope of a municipality’s immunity from liability under § 1983 is essentially one of statutory construction, see Wood v. Strickland, 420 U. S. 308, 314, 316 (1975); Tenney v. Brandhove, 341 U. S. 367, 376 (1951), the starting point in our analysis must be the language of the statute itself. Andrus v. Allard, 444 U. S. 51, 56 (1979); Blue Chip Stamps v. Manor Drug Stores, 421 U. S. 723, 756 (1975) (Powell, J., concurring). By its terms, § 1983 “creates a species of tort liability that on its face admits of no immunities.” Imbler v. Pachtman, 424 U. S. 409, 417 (1976). Its language is absolute and unqualified; no mention is made of any privileges, immunities, or defenses that may be asserted. Bather, the Act imposes liability upon “every person” who, under color of state law or custom, “subjects, or causes to be subjected, any citizen of the United States ... to the deprivation of any rights, privileges, or immunities secured by the Constitution and laws.”16 And Monell held that these words were intended to encompass municipal corporations as well as natural “persons.”

Moreover, the congressional debates surrounding the passage of § 1 of the Civil Rights Act of 1871, 17 Stat. 13 — the forerunner of § 1983 — confirm the expansive sweep of the stat*636utory language. Representative Shellabarger, the author and manager of the bill in the House, explained in his introductory remarks the breadth of construction that the Act was to receive:

“I have a single remark to make in regard to the rule of interpretation of those provisions of the Constitution under which all the sections of the bill are framed. This act is remedial, and in aid of the preservation of human liberty and human rights. All statutes and constitutional provisions authorizing such statutes are liberally and beneficently construed. It would be most strange and, in civilized law, monstrous were this not the rule of interpretation. As has been again and again decided by your own Supreme Court of the United States, and everywhere else where there is wise judicial interpretation, the largest latitude consistent with the words employed is uniformly given in construing such statutes and constitutional provisions as are meant to protect and defend and give remedies for their wrongs to all the people.” Cong. Globe, 42d Cong., 1st Sess., App. 68 (1871) (hereinafter Globe App.).

Similar views of the Act’s broad remedy for violations of federally protected rights were voiced by its supporters in both Houses of Congress. See Monell v. New York City Dept. of Social Services, 436 U. S., at 683-687.17

*637However, notwithstanding § 1983’s expansive language and the absence of any express incorporation of common-law immunities, we have, on several occasions, found that a tradition of immunity was so firmly rooted in the common law and was supported by such strong policy reasons that “Congress would have specifically so provided had it wished to abolish the doctrine.” Pierson v. Ray, 386 U. S. 547, 555 (1967). Thus in Tenney v. Brandhove, supra, after tracing the development of an absolute legislative privilege from its source in 16th-century England to its inclusion in the Federal and State Constitutions, we concluded that Congress “would [not] impinge on a tradition so well grounded in history and reason by covert inclusion in the general language” of § 1983. 341 U. S., at 376.

Subsequent cases have required that we consider the personal liability of various other types of government officials. Noting that “[f]ew doctrines were more solidly established at common law than the immunity of judges from liability for damages for acts committed within their judicial jurisdiction,” Pierson v. Ray, supra, at 553-554, held that the absolute immunity traditionally accorded judges was preserved under § 1983. In that same case, local police officers were held to enjoy a “good faith and probable cause” defense to § 1983 suits similar to that which existed in false arrest actions at common law. 386 U. S., at 555-557. Several more recent decisions have found immunities of varying scope appropriate for different state and local officials sued under § 1983. See Procunier v. Navarette, 434 U. S. 555 (1978) (qualified im*638munity for prison officials and officers); Imbler v. Pachtman, 424 U. S. 409 (1976) (absolute immunity for prosecutors in initiating and presenting the State’s case); O’Connor v. Donaldson, 422 U. S. 563 (1975) (qualified immunity for superintendent of state hospital); Wood v. Strickland, 420 U. S. 308 (1975) (qualified immunity for local school board members) ; Scheuer v. Rhodes, 416 U. S. 232 (1974) (qualified “good-faith” immunity for state Governor and other executive officers for discretionary acts performed in the course of official conduct).

In each of these cases, our finding of § 1983 immunity “was predicated upon a considered inquiry into the immunity historically accorded the relevant official at common law and the interests behind it.” Imbler v. Pachtman, supra, at 421. Where the immunity claimed by the defendant was well established at common law at the time § 1983 was enacted, and where its rationale was compatible with the purposes of the Civil Eights Act, we have construed the statute to incorporate that immunity. But there is no tradition of immunity for municipal corporations, and neither history nor policy supports a construction of § 1983 that would justify the qualified immunity accorded the city of Independence by the Court of Appeals. We hold, therefore, that the municipality may not assert the good faith of its officers or agents as a defense to liability under § 1983.18

A

Since colonial times, a distinct feature of our Nation’s system of governance has been the conferral of political power upon public and municipal corporations for the management of matters of local concern. As Monell recounted, by 1871, *639municipalities — like private corporations — were treated as natural persons for virtually all purposes of constitutional and statutory analysis. In particular, they were routinely sued in both federal and state courts. See 436 U. S., at 687-688. Cf. Cowles v. Mercer County, 7 Wall. 118 (1869). Local governmental units were regularly held to , answer in damages for a wide range of statutory and constitutional violations, as well as for common-law actions for breach of contract.19 And although, as we discuss below,20 a municipal*640ity was not subject to suit for all manner of tortious conduct, it is clear that at the time § 1983 was enacted, local governmental bodies did not enjoy the sort of “good-faith” qualified immunity extended to them by the Court of Appeals.

As a general rule, it was understood that a municipality’s tort liability in damages was identical to that of private corporations and individuals:

“There is nothing in the character of a municipal corporation which entitles it to an immunity from liability for such malfeasances as private corporations or individuals would be liable for in a civil action. A municipal corporation is liable to the same extent as an individual for any act done by the express authority of the corporation, or of a branch of its government, empowered to act for it upon the subject to which the particular act relates, and for any act which, after it has been done, has been lawfully ratified by the corporation.” T. Shear-man & A. Redfield, A Treatise on the Law of Negligence § 120, p. 139 (1869) (hereinafter Shearman & Redfield).

Accord, 2 Dillon § 764, at 875 (“But as respects municipal corporations proper, ... it is, we think, universally considered, even in the absence of statute giving the action, that they are liable for acts of misfeasance positively injurious to individuals, done by their authorized agents or officers, in the course of the performance of corporate powers constitutionally conferred, or in the execution of corporate duties”) (emphasis in original). See 18 E. McQuillin, Municipal Corporations § 53.02 (3d rev. ed. 1977) (hereinafter McQuillin). Under this general theory of liability, a municipality was deemed responsible for any private losses generated through a wide variety of its operations and functions, from personal injuries due to its defective sewers, thoroughfares, and public utilities, to property damage caused by its trespasses and uncompensated takings.21

*641Yet in the hundreds of cases from that era awarding damages against municipal governments for wrongs committed by them, one searches in vain for much mention of a qualified immunity based on the good faith of municipal officers. Indeed, where the issue was discussed at all, the courts had rejected the proposition that a municipality should be privileged where it reasonably believed its actions to be lawful. In the leading case of Thayer v. Boston, 36 Mass. 511, 515-516 (1837), for example, Chief Justice Shaw explained:

“There is a large class of cases, in which the rights of both the public and of individuals may be deeply involved, in which it cannot be known at the time the act is done, whether it is lawful or not. The event of a legal inquiry, in a court of justice, may show that it was unlawful. Still, if it was not known and understood to be! unlawful at the time, if it was an act done by the officers! having competent authority, either by express vote of \ the city government, or by the nature of the duties and ! functions with which they are charged, by their offices, to act upon the general subject matter, and especially if the j act was done with an honest view to obtain for the public j some lawful benefit or advantage, reason and justice ob- ¡ viously require that the city, in its corporate capacity, should be liable to make good the damage sustained by an individual, in consequence of the acts thus done.” ]

The Thayer principle was later reiterated by courts in several jurisdictions, and numerous decisions awarded damages against municipalities for violations expressly found to have been committed in good faith. See, e. g., Town Council of Akron v. McComb, 18 Ohio 229, 230-231 (1849); Horton v. Inhabitants of Ipswich, 66 Mass. 488, 489, 492 (1853); Elliot v. Concord, 27 N. H. 204 (1853); Hurley v. Town of Texas, 20 Wis. 634, 637-638 (1866); Lee v. Village of Sandy Hill, 40 N. Y. *642442, 448-451 (1869); Billings v. Worcester, 102 Mass. 329, 332-333 (1869); Squiers v. Village of Neenah, 24 Wis. 588, 593 (1869); Hawks v. Charlemont, 107 Mass. 414, 417-418 (1871).22

That municipal corporations were commonly held liable for damages in tort was also recognized by the 42d Congress. See Monell v. New York City Dept. of Social Services, 436 U. S., at 688. For example, Senator Stevenson, in opposing the Sherman amendment’s creation of a municipal liability for the riotous acts of its inhabitants, stated the prevailing law: "Numberless cases are to be found where a statutory liability has been created against municipal corporations for injuries resulting from a neglect of corporate duty.” Cong. *643Globe, 42d Cong., 1st Sess., 762 (hereinafter Globe).23 Nowhere in the debates, however, is there a suggestion that the common law excused a city from liability on account of the good faith of its authorized agents, much less an indication of a congressional intent to incorporate such an immunity into the Civil Rights Act.24 The absence of any allusion to a municipal immunity assumes added significance in light of the objections raised by the opponents of § 1 of the Act that its unqualified language could be interpreted to abolish the traditional good-faith immunities enjoyed by legislators, judges, governors, sheriffs, and other public officers.25 Had *644there been a similar common-law immunity for municipalities, the bill’s opponents doubtless would have raised the specter of its destruction, as well.

To be sure, there were two doctrines that afforded municipal corporations some measure of protection from tort liability. The first sought to distinguish betweeá a municipality’s “governmental” and “proprietary” functions; as to the former, the city was held immune, whereas in its exercise of the latter, the city was held to the same standards of liability as any private corporation. The second doctrine immunized a municipality for its “discretionary” or “legislative” activities, but not for those which were “ministerial” in nature. A brief examination of the application and the rationale underlying each of these doctrines demonstrates that Congress could not have intended them to limit a municipality’s liability under § 1983.

The governmental-proprietary distinction 26 owed its existence to the dual nature of the municipal corporation. On *645the one hand, the municipality was a corporate body, capable of performing the same “proprietary” functions as any private corporation, and liable for its torts in the same manner and to the same extent, as well. On the other hand, the municipality was an arm of the State, and when acting in that “governmental” or “public” capacity, it shared the immunity traditionally accorded the sovereign.27 But the principle of sovereign immunity — itself a somewhat arid fountainhead for municipal immunity28 — is necessarily nullified when the *646State expressly or impliedly allows itself, or its creation, to be sued. Municipalities were therefore liable not only for their “proprietary” acts, but also for those “governmental” functions as to which the State had withdrawn their immunity. And, by the end of the 19th century, courts regularly held that in imposing a specific duty on the municipality either in its charter or by statute, the State had impliedly withdrawn the city’s immunity from liability for the nonperformance or misperformance of its obligation. See, e. g., Weightman v. The Corporation of Washington, 1 Black 39, 50-52 (1862); Providence v. Clapp, 17 How. 161, 167-169 (1855). See generally Shearman & Redfield §§ 122-126; Note, Liability of Cities for the Negligence and Other Misconduct of their Officers and Agents, 30 Am. St. Rep. 376, 385 (1893). Thus, despite the nominal existence of an immunity for “governmental” functions, municipalities were found *647liable in damages in a multitude of cases involving such activities.

That the municipality’s common-law immunity for “governmental” functions derives from the principle of sovereign immunity also explains why that doctrine could not have served as the basis for the qualified privilege respondent city claims under § 1983. First, because, sovereign immunity insulates the municipality from unconsented suits altogether, the pres-enee or absence of good faith is simply irrelevant. The critical issue is whether injury occurred while the city was exercising- governmental, as opposed to pioprietary, powers or obligations — not whether its agents reasonably believed they were acting lawfully in so conducting themselves.29 Morfundamentally, however, the municipality’s “governmental” immunity is obviously abrogated by the sovereign’s enacment of a statute making it amenable to suit. Section 1983 was just such a statute. By including municipalities within the class of “persons” subject to liability for violations of the Federal Constitution and laws, Congress — the supreme sovereign on matters of federal law30 — abolished whatever ves*648tige of the State’s sovereign immunity the municipality possessed.

The second common-law distinction between municipal functions — that protecting the city from suits challenging “discretionary” decisions — was grounded not on the principle of sovereign immunity, but on a concern for separation of powers. A large part of the municipality’s responsibilities involved broad discretionary decisions on issues of public policy — decisions that affected large numbers of persons and called for a delicate balancing of competing considerations. For a court or jury, in the guise of a tort suit, to review the reasonableness of the city’s judgment on these matters would be an infringement upon the powers properly vested in a coordinate and coequal branch of government. See Johnson v. State, 69 Cal. 2d 782, 794, n. 8, 447 P. 2d 352, 361, n. 8 (1968) (en banc) (“Immunity for ‘discretionary’ activities serves no purpose except to assure that courts refuse to pass judgment on policy decisions in the province of coordinate branches of government”). In order to ensure against any invasion into the legitimate sphere of the municipality’s policymaking processes, courts therefore refused to entertain suits against the city “either for the non-exercise of, or for the manner in which in good faith it exercises, discretionary powers of a public or legislative character.” 2 Dillon § 753, at 862.31

Although many, if not all, of a municipality’s activities would seem to involve at least some measure of discretion, the influence of this doctrine on the city’s liability was not as significant as might be expected. For just as the courts *649implied an exception to the municipality’s immunity for its “governmental” functions, here, too, a distinction was made that had the effect of subjecting the city to liability for much of its tortious conduct. While the city retained its immunity for decisions as to whether the public interest required acting in one manner or another, once any particular decision was made, the city was fully liable for any injuries incurred in the execution of its judgment. See, e. g., Hill v. Boston, 122 Mass. 344, 358-359 (1877) (dicta) (municipality would be immune from liability for damages resulting from its decision where to construct sewers, since that involved a discretionary judgment as to the general public interest; but city would be liable for neglect in the construction or repair of any particular sewer, as such activity is ministerial in nature). See generally C. Rhyne, Municipal Law § 30.4, pp. 736-737 (1957); Williams § 7. Thus municipalities remained liable in damages for a broad range of conduct implementing their disere-/ tionary decisions.

Once again, an understanding of the rationale underlying the common-law immunity for “discretionary” functions explains why that doctrine cannot serve as the foundation for a good-faith immunity under § 1983. That common-law doctrine merely prevented courts from substituting their own judgment on matters within the lawful discretion of the municipality. But a municipality has no “discretion” to violate the Federal Constitution; its dictates are absolute and imperative. And when a court passes judgment on the municipality’s conduct in a § 1983 action, it does not seek to second-guess the “reasonableness” of the city’s decision nor to interfere with the local government’s resolution of competing policy considerations. Rather, it looks only to whether the municipality has conformed to the requirements of the Federal Constitution and statutes. As was stated in Sterling v. Constantin, 287 U. S. 378, 398 (1932): “When there is a substantial showing that the exertion of state power has *650overridden private rights secured by that Constitution, the subject is necessarily one for judicial inquiry in an appropriate proceeding directed against the individuals charged with the transgression.”

In sum, we can discern no “tradition so well grounded in history and-reason” that would warrant the conclusion that in enacting § 1 of the Civil Rights Act, the 42d Congress sub silentio extended to municipalities a qualified immunity based on the good faith of their officers. Absent any clearer indication that Congress intended so to limit the reach of a statute expressly designed to provide a “broad remedy for violations of federally protected civil rights,” Monell v. New York City Dept. of Social Services, 436 U. S., at 685, we are unwilling to suppose that injuries occasioned by a municipality’s unconstitutional conduct were not also meant to be fully redressable through its sweep.32

B

Our rejection of a construction of § 1983 that would accord municipalities a qualified immunity for their good-faith constitutional violations is compelled both by the legislative purpose in enacting the statute and by considerations of public policy. The central aim of the Civil Rights Act was to provide protection to those persons wronged by the “ ‘[mjisuse of power, possessed by virtue of state law and made possible only because the wrongdoer is clothed with the authority of state law.’ ” Monroe v. Pape, 365 U. S., at 184 (quoting United States v. Classic, 313 U. S. 299, 326 (1941)). By creating an express federal remedy, Congress sought to “enforce provisions of the Fourteenth Amendment against those *651who carry a badge of authority of a State and represent it in some capacity, whether they act in accordance with their authority or misuse it.” Monroe v. Pape, supra, at 172.

How “uniquely amiss” it would be, therefore, if the government itself — “the social organ to which all in our society look for the promotion of liberty, justice, fair and equal treatment, and the setting of worthy norms and goals for social conduct” — were permitted to disavow liability for the injury it has begotten. See Adickes v. Kress & Co., 398 U. S. 144, 190 (1970) (opinion of Brennan, J.). A damages remedy against the offending' party is a vital component of any scheme for vindicating cherished constitutional guarantees, and the importance of assuring its efficacy is only accentuated when the wrongdoer is the institution that has been established to protect the very rights it has transgressed: Yet owing to the qualified immunity enjoyed by most government officials, see Scheuer v. Rhodes, 416 U. S. 232 (1974), many victims of municipal malfeasance would be left remediless if the city were also allowed to assert a good-faith defense. Unless countervailing considerations counsel otherwise, the injustice of such a result should not be tolerated.33

Moreover, § 1983 was intended not only to provide compensation to the victims of past abuses, but to serve as a deterrent against future constitutional deprivations, as well. See Robertson v. Wegmann, 436 U. S. 584, 590-591 (1978); Carey v. Piphus, 435 U. S. 247, 256-257 (1978). The knowledge that a municipality will be liable for all of its injurious conduct, whether committed in good faith or not, should create *652an incentive for officials who may harbor doubts about the lawfulness of their intended actions to err on the side of protecting citizens’ constitutional rights.34 Furthermore, the threat that damages might be levied against the city may encourage those in a policymaking position to institute internal rules and programs designed to minimize the likelihood of unintentional infringements on constitutional rights.35 Such procedures are particularly beneficial in preventing those “systemic” injuries that result not so much from the conduct of any single individual, but from the interactive behavior of several government officials, each of whom may be acting in good faith. Cf. Note, Developments in the Law: Section 1983 and Federalism, 90 Harv. L. Rev. 1133, 1218-1219 (1977).36

Our previous decisions conferring qualified immunities on various government officials, see supra, at 637-638, are not to *653be read as derogating the significance of the societal interest in compensating the innocent victims of governmental misconduct. Rather, in each case we concluded that overriding considerations of public policy nonetheless demanded that the official be given a measure of protection from personal liability. The concerns that justified those decisions, however, are less compelling, if not wholly inapplicable, when the liability of the municipal entity is at issue.37

*654In Scheuer v. Rhodes, supra, at 240, The Chief Justice identified the two “mutually dependent rationales” on which the doctrine of official immunity rested:

“(1) the injustice, particularly in the absence of bad faith, of subjecting to liability an officer who is required, by the legal obligations of his position, to exercise discretion; (2) the danger that the threat of such liability would deter his willingness to execute his office with the decisiveness and the judgment required by the public good.”38

The first consideration is simply not implicated when the damages award comes not from the official’s pocket, but from the public treasury. It hardly seems unjust to require a municipal defendant which has violated a citizen’s constitutional rights to compensate him for the injury suffered thereby. Indeed, Congress enacted § 1983 precisely to provide a remedy for such abuses of official power. See Monroe v. Pape, 365 U. S., at 171-172. Elemental notions of fairness dictate that one who causes a loss should bear the loss.

It has been argued, however, that revenue raised by taxation for public use should not be diverted to the benefit of a single or discrete group of taxpayers, particularly where the municipality has at all times acted in good faith. On the contrary, the accepted view is that stated in Thayer v. Boston — " that the city, in its corporate capacity, should be liable to make good the damage sustained by an [unlucky] indi*655vidual, in consequence of the acts thus done.” 36 Mass., at 515. After all, it is the public at large which enjoys the benefits of the government’s activities, and it is the public at large which is ultimately responsible for its administration. Thus, even where some constitutional development could not have been foreseen by municipal officials, it is fairer to allocate any resulting financial loss to the inevitable costs of government borne by all the taxpayers, than to allow its impact to be felt solely by those whose rights, albeit newly recognized, have been violated. See generally 3 K. Davis, Administrative Law Treatise §25.17 (1958 and Supp. 1970); Prosser § 131, at 978; Michelman, Property, Utility, and Fairness: Some Thoughts on the Ethical Foundations of “Just Compensation” Law, 80 Harv. L. Rev. 1165 (1967).39

The second rationale mentioned in Scheuer also loses its force when it is the municipality, in contrast to the official, whose liability is at issue. At the heart of this justification for a qualified immunity for the individual official is the concern that the threat of personal monetary liability will introduce an unwarranted and unconscionable consideration into the decisionmaking process, thus paralyzing the governing official’s decisiveness and distorting his judgment on matters *656of public policy.40 The inhibiting effect is significantly reduced, if not eliminated, however, when the threat of personal liability is removed. First, as an empirical matter, it is questionable whether the hazard of municipal loss will deter a public officer from the conscientious exercise of his duties; city officials routinely make decisions that either require a large expenditure of municipal funds or involve a substantial risk of depleting the public fisc. See Kostka v. Hogg, 560 F. 2d 37, 41 (CA1 1977). More important, though, is the realization that consideration of the municipality’s liability for constitutional violations is quite properly the concern of its elected or appointed officials. Indeed, a decisionmaker would be derelict in his duties if, at some point, he did not consider whether his decision comports with constitutional mandates and did not weigh the risk that a violation might result in an award of damages from the public treasury. As one commentator aptly put it: “Whatever other concerns should shape a particular official's actions, certainly one of them should be the constitutional rights of individuals who will be affected by his actions. To criticize section 1983 liability because it leads decisionmakers to avoid the infringement of constitutional rights is to criticize one of the statute’s raisons d’etre.”41

*657IV

In sum, our decision holding that municipalities have no immunity from damages liability flowing from their constitutional violations harmonizes well with developments in the common law and our own pronouncements on official immunities under § 1983. Doctrines of tort law have changed significantly over the past century, and our notions of governmental responsibility should properly reflect that evolution. No longer is individual “blameworthiness” the acid test of liability; the principle of equitable loss-spreading has.joined fault as a factor in distributing the costs of official misconduct.

We believe that today’s decision, together with prior precedents in this area, properly allocates these costs among the three principals in the scenario of the § 1983 cause of action: the victim of the constitutional deprivation; the officer whose conduct caused the injury; and the public, as represented by the municipal entity. The innocent individual who is harmed by an abuse of governmental authority is assured that he will be compensated for his injury. The offending official, so long as he conducts himself in good faith, may go about his business secure in the knowledge that a qualified immunity will protect him from personal liability for damages that are more appropriately chargeable to the populace as a whole. And the public will be forced to bear only the costs of injury inflicted by the “execution of a government’s policy or custom, whether made by its lawmakers or by those whose edicts or acts may fairly be said to represent official policy.” *658Monell v. New York City Dept. of Social Services, 436 U. S., at 694.

Reversed.

Mr. Justice Powell.,

with whom The Chief Justice, Mr. Justice Stewart, and Mr. Justice Rehnquist join, dissenting.

The Court today holds that the city of Independence may be liable in damages for violating a constitutional right that was unknown when the events in this case, occurred. It finds a denial of due process in the city’s failure to grant petitioner a hearing to clear his name after he was discharged. But his dismissal involved only the proper exercise of discretionary powers according to prevailing constitutional doctrine. The city imposed no stigma on petitioner that would require a “name clearing” hearing under the Due Process Clause.

On the basis of this alleged deprivation of rights, the Court interprets 42 U. S. C. § 1983 to impose strict liability on municipalities for constitutional violations. This strict liability approach inexplicably departs from this Court’s prior decisions under § 1983 and runs counter to the concerns of the 42d Congress when it enacted the statute. The Court’s ruling also ignores the vast weight of common-law precedent as well as the current state law of municipal immunity. For these reasons, and because this decision will hamper local governments unnecessarily, I dissent.

I

The Court does not question the District Court’s statement of the facts surrounding Owen’s dismissal. Ante, at 625. It nevertheless rejects the District Court’s conclusion that no due process hearing was necessary because “the circumstances of [Owen’s] discharge did not impose a stigma of illegal or immoral conduct on his professional reputation.” 421 F. Supp. 1110, 1122 (WD Mo. 1976); see ante, at 633-634, n. 13. *659Careful analysis of the record supports the District Court’s view that Owen suffered no constitutional deprivation.

A

From 1967 to 1972, petitioner Owen served as Chief of the Independence Police Department at the pleasure of the City Manager.1 Friction between Owen and City Manager Alberg flared openly in early 1972, when charges surfaced that the Police Department’s property room was mismanaged. The City Manager initiated a full internal investigation.

In early April, the City Auditor reported that the records in the property room were so sparse that he could not conduct an audit. The City Counselor reported that “there was no evidence of any criminal acts, or violation of any state law or municipal ordinances, in the administration of the property room.” 560 F. 2d 925, 928 (CA8 1977). In a telephone call on April 10, the City Manager asked Owen to resign and offered him another position in the Department. The two met on the following day. Alberg expressed his unhappiness over the property room situation and again requested that Owen step down. When Owen refused, the City Manager responded that he would be fired. 421 F. Supp., at 1114-1115.

On April 13, the City Manager asked Lieutenant Cook of the Police Department if he would be willing to take over as Chief. Alberg also released the following statement to the public:

“At my direction, the City Counselor’s office, [i]n conjunction with the City Auditor ha[s] ¡completed a routine audit of the police property room.

*660“Discrepancies were found in the administration, handling and security of recovered property. There appears to be no evidence to substantiate any allegations of a criminal nature. . . .” 560 F. 2d, at 928-929.

The District Court found that the City Manager decided on Saturday, April 15, to replace Owen with Lieutenant Cook as Chief of Police. 421 F. Supp., at 1115. Before the decision was announced, however, City Council Member Paul Roberts obtained the internal reports on the property room. At the April 17 Council meeting, Roberts read a prepared statement that accused police officials of “gross inefficiencies” and various “inappropriate” actions. Id., at 1116, n. 2. He then moved that the Council release the reports to the public, refer them to the Prosecuting Attorney of Jackson County for presentation to a grand jury, and recommend to the City Manager that he “take all direct and appropriate action permitted under the Charter. . . .” Ibid. The Council unanimously approved the resolution.

On April 18, Alberg “implemented his prior decision to discharge [Owen] as Chief of Police.” 560 F. 2d, at 929. The notice of termination stated simply that Owen’s employment was “[t]erminated under the provisions of Section 3.3 (1) of the City Charter.” App. 17. That charter provision grants the City Manager complete authority to remove “directors” of administrative departments “when deemed necessary for the good of the service.” Owen’s lawyer requested a hearing on his client’s termination. The Assistant City Counselor responded that “there is no appellate procedure or forum provided by the Charter or ordinances of the City of Independence, Missouri, relating to the dismissal of Mr. Owen.” Id., at 27.

The City Manager referred to the Prosecuting Attorney all reports on the property room. The grand jury returned a “no true bill,” and there has been no further official action on the matter. Owen filed a state lawsuit against Councilman *661Roberts and City Manager Alberg, asking for damages for libel, slander, and malicious prosecution. Alberg won a dismissal of the state-law claims against him, and Councilman Roberts reached a settlement with Owen.2

This federal action was filed in 1976. Owen alleged that he was denied his liberty interest in his professional reputation when he was dismissed without formal charges or a hearing. Id., at 8, 10.3

B

Due process requires a hearing on the discharge of a government employee “if the employer creates and disseminates a false and defamatory impression about the employee in connection with his termination. . . .” Codd v. Velger, 429 U. S. 624, 628 (1977) (per curiam). This principle was first announced in Board of Regents v. Roth, 408 U. S. 564 (1972), which was decided in June 1972, 10 weeks after Owen was discharged. The pivotal question after Roth is whether the circumstances of the discharge so blackened the employee’s *662name as to impair his liberty interest in his professional reputation. Id., at 572-575.

The events surrounding Owen’s dismissal “were prominently reported in local newspapers.” 560 F. 2d, at 930. Doubtless, the public received a negative impression of Owen’s abilities and performance. But a “name clearing” hearing is not necessary unless the employer makes a public statement that “might seriously damage [the employee’s] standing and associations in his community.” Board of Regents v. Roth, supra, at 573. No hearing is required after the “discharge of a public employee whose position is terminable at the will of the employer when there is no public disclosure of the reasons for the discharge.” Bishop v. Wood, 426 U. S. 341, 348 (1976).

The City Manager gave no specific reason for dismissing Owen. Instead, he relied on his discretionary authority to discharge top administrators “for the good of the service.” Alberg did not suggest that Owen “had been guilty of dishonesty, or immorality.” Board of Regents v. Roth, supra, at 573. Indeed, in his “property room” statement of April 13, Alberg said that there was “no evidence to substantiate any allegations of a criminal nature.” This exoneration was reinforced by the grand jury’s refusal to initiate a prosecution in the matter. Thus, nothing in the actual firing cast such a stigma on Owen’s professional reputation that his liberty was infringed.

The Court does not address directly the question whether any stigma was imposed by the discharge. Rather, it relies on the Court of Appeals’ finding that stigma derived from events “connected with” the firing. Ante, at 633; 589 F. 2d, at 337. That court attached great significance to the resolution adopted by the City Council at its April 17 meeting. But the resolution merely recommended that Alberg take “appropriate action,” and the District Court found no “causal connection” between events in the City Council and the firing of Owen. 421 F. Supp., at 1121. Two days *663before the Council met, Alberg already had decided to dismiss Owen. Indeed, Councilman Roberts stated at the meeting that the City Manager had asked for 'Owen’s resignation. Id., at 1116, n. 2.4

Even if the Council resolution is viewed as part of the discharge process, Owen has demonstrated no denial of his liberty; Neither the City Manager nor the Council cast any aspersions on Owen’s character. Alberg absolved all connected with the property rodm of any illegal activity, while the Council resolution alleged no wrongdoing. That events focused public attention upon Owen’s dismissal is undeniable;' such attention is a condition of employment — and of discharge — for high government officials. Nevertheless, nothing in the actions of the City Manager or the City Council triggered a constitutional right to a name-clearing hearing.5

The statements by Councilman Roberts were neither measured nor benign, but they provide no basis for this action against the city of Independence. Under Monell v. New York City Dept. of Social Services, 436 U. S. 658, 691 (1978), the city cannot be held liable for Roberts’ statements on a theory of respondeat superior. That case held that § 1983 *664makes municipalities liable for constitutional deprivations only if the challenged action was taken “pursuant to official municipal policy of some nature. . . As the Court noted, “a municipality cannot be held liable solely because it employs a tortfeasor. . . .” 436 U. S., at 691 (emphasis in original). The statements of a single councilman scarcely rise to the level of municipal policy.6

As the District Court concluded, “[a]t most, 'the circumstances . . . suggested that, as Chief of Police, [Owen] had been an inefficient administrator.” 421 F. Supp., at 1122. This Court now finds unconstitutional stigma in the interaction of unobjectionable official acts with the unauthorized statements of a lone councilman who had no direct role in the discharge process. The notoriety that attended Owen’s firing resulted not from any city policy, but solely from public misapprehension of the reasons for a purely discretionary dismissal. There was no constitutional injury; there should be no liability.7

II

Having constructed a constitutional deprivation from the valid exercise of governmental authority, the Court holds that municipalities are strictly liable for their constitutional torts. Until two years ago, municipal corporations enjoyed absolute immunity from § 1983 claims. Monroe v. Pape, 365 U. S. *665167 (1961). But Monell v. New York City Dept. of Social Services, supra, held that local governments are “persons” within the meaning of the statute, and thus are liable in damages for constitutional violations inflicted by municipal policies. 436 U. S., at 690. Monell did not address the question whether municipalities might enjoy a qualified immunity or good-faith defense against § 1983 actions. 436 U. S., at 695, 701; id., at 713-714 (Powell, J., concurring).

After today’s decision, municipalities will have gone in two short years from absolute immunity under § 1983 to strict liability. As a policy matter, I believe that strict municipal liability unreasonably subjects local governments to damages judgments for actions that were reasonable when performed. It converts municipal governance into a hazardous slalom through constitutional obstacles that often are unknown and unknowable.

The Court’s decision also impinges seriously on the prerogatives of municipal entities created and regulated primarily by the States. At the very least, this Court should not initiate a federal intrusion of this magnitude in the absence of explicit congressional action. Yet today’s decision is supported by nothing in the text of § 1983. Indeed, it conflicts with the apparent intent of the drafters of the statute, with the common law of municipal tort liability, and with the current state law of municipal immunities.

A

1

Section 1983 provides a private right of action against “[e]very person” acting under color of state law who imposes or causes to be imposed a deprivation of constitutional rights.8 *666Although the statute does not refer to immunities, this Court has held that the law “is to be read in harmony with general principles of tort immunities and defenses rather than in derogation of them.” Imbler v. Pachtman, 424 U. S. 409, 418 (1976); see Tenney v. Brandhove, 341 U. S. 367, 376 (1951).

This approach reflects several concerns. First, the common-law traditions of immunity for public officials could not have been repealed by the “general language” of § 1983. Tenney v. Brandhove, supra, at 376; see Imbler v. Pachtman, supra, at 421-424; Pierson v. Ray, 386 U. S. 547, 554-555 (1967). In addition, “the public interest requires decisions and action to enforce laws for the protection of the public.” Scheuer v. Rhodes, 416 U. S. 232, 241 (1974). Because public officials will err at times, “[t]he concept of immunity assumes . . . that it is better to risk some error and possibly injury from such error than not to decide or act at all.” Id., at 242; see Wood v. Strickland, 420 U. S. 308, 319-320 (1975). By granting some immunity to governmental actors, the Court has attempted to ensure that public decisions will not be dominated by fears of liability for actions that may turn out to be unconstitutional. Public officials “cannot be expected to predict the future course of constitutional law. . . .” Procunier v. Navarette, 434 U. S. 555, 562 (1978).

In response to these considerations, the Court has found absolute immunity from § 1983 suits for state legislators, Tenney v. Brandhove, supra, judges, Pierson v. Ray, supra, at 553-555, and prosecutors in their role as advocates for the State, Imbler v. Pachtman, supra. Other officials have been granted a qualified immunity that protects them when in good faith they have implemented policies that reasonably were thought to be constitutional. This limited immunity extends to police officers, Pierson v. Ray, supra, at 555-558, state executive officers, Scheuer v. Rhodes, supra, local school board members, Wood v. Strickland, supra, the super*667intendent of a state hospital, O’Connor v. Donaldson, 422 U. S. 563, 576-577 (1975), and prison officials, Procunier v. Navarette, supra.

The Court today abandons any attempt to harmonize § 1983 with traditional tort law. ' It points out that municipal immunity may be abrogated by legislation. Thus, according to the Court, Congress “abolished” municipal immunity when it included municipalities “within the class of ‘persons’ subject to liability” under § 1983. Ante, at 647.

This reasoning flies in the face of our prior decisions under this statute. We. have held repeatedly that “immunities ‘well grounded in history and reason’ [were not] abrogated ‘by covert inclusion in the general language’ of § 1983.” Imbler v. Pachtman, supra, at 418, quoting Tenney v. Brandhove, supra, at 376. See Scheuer v. Rhodes, supra, at 243-244; Pierson v. Ray, supra, at 554. The peculiar nature of the Court’s position emerges when the status of executive officers under § 1983 is compared with that of local governments. State and local executives are personally liable for bad-faith or unreasonable constitutional torts. Although Congress' had the power to make those individuals liable for all such torts, this Court has refused to find an abrogation of traditional immunity in a statute that does not mention immunities. Yet the Court now views the enactment of § 1983 as a direct abolition of traditional municipal immunities. Unless the Court is overruling its previous immunity decisions, the silence in § 1983 must mean that the 42d Congress- mutely accepted the immunity of executive officers, but silently rejected common-law municipal immunity. I find this interpretation of the statute singularly implausible.

2

Important public policies support the extension of qualified immunity to local governments. First, as recognized by the doctrine of separation of powers, some governmental decisions should be at least presumptively insulated from judicial re*668view. Mr. Chief Justice Marshall wrote in Marbury v. Madison, 1 Cranch 137, 170 (1803), that “[t]he province of the court is . . . not to inquire how the executive, or executive officers, perform duties in which they have a discretion.” Marshall stressed the caution with which courts must approach “[qjuestions, in their nature political, or which are, by the constitution and laws, submitted to the executive.” The allocation of public resources and the operational policies of the government itself are activities that lie peculiarly within the competence of executive and legislative bodies. When charting those policies, a local official should not have to gauge his employer’s possible liability under § 1983 if he incorrectly — though reasonably and in good faith — forecasts the course"of constitutional law. Excessive judicial intrusion into such decisions can only distort municipal decisionmaking and discredit the courts. Qualified immunity would provide presumptive protection for discretionary acts, while still leaving the municipality liable for bad faith or unreasonable constitutional deprivations.

Because today’s decision will inject constant consideration of § 1983 liability into local decisionmaking, it may restrict the independence of local governments and their ability to respond to the needs of their communities. Only this Term, we noted that the “point” of immunity under § 1983 “is to forestall an atmosphere of intimidation that would conflict with [officials’] resolve to perform their designated functions in a principled fashion.” Ferri v. Ackerman, 444 U. S. 193, 203-204 (1979).

The Court now argues that local officials might modify their actions unduly if they face personal liability under § 1983, but that they are unlikely to do so when the locality itself will be held liable. Ante, at 655-656. This contention denigrates the sense of responsibility of municipal officers, and misunderstands the political process. Responsible local officials will be concerned about potential judgments against *669their municipalities for alleged constitutional torts. Moreover, they will be accountable within the political system for subjecting the municipality to adverse judgments. If officials must look over their shoulders at strict municipal liability for unknowable constitutional deprivations, the resulting degree of governmental paralysis will be little different from that caused by fear of personal liability. Cf. Wood v. Strickland, 420 U. S., at 319-320; Scheuer v. Rhodes, 416 U. S., at 242.9

In addition, basic fairness requires a qualified immunity for municipalities. The good-faith defense recognized under § 1983 authorizes liability only when officials acted with malicious intent or when they “knew or should have known that their conduct violated the constitutional norm.” Procunier v. Navarette, 434 U. S., at 562. The standard.incorporates the idea that liability should not attach unless there was notice that a constitutional right was at risk. This idea applies to governmental entities and individual officials alike. Constitutional law is what the courts say it is, and — as demonstrated by today’s decision and its precursor, Monell — even the most prescient lawyer would hesitate to give a firm opinion on matters not plainly settled. Municipalities, often acting in the utmost good faith, may not know or anticipate when their action or inaction will be deemed a constitutional violation.10

*670The Court nevertheless suggests that, as a matter of social justice, municipal corporations should be strictly liable even if they could not have known that a particular action would violate the Constitution. After all, the Court urges, local governments can “spread” the costs of any judgment across the local population. Ante, at 655. The Court neglects, however, the fact that many local governments lack the resources to withstand substantial unanticipated liability under § 1983. Even enthusiastic proponents of municipal liability have conceded that ruinous judgments under the statute could imperil local governments. E. g., Note, Damage Remedies Against Municipalities for Constitutional Violations, 89 Harv. L. Rev. 922, 958 (1976).11 By simplistically applying the theorems of welfare economics and ignoring the reality of municipal finance, the Court imposes strict liability on the level of government least able to bear it.12 For some municipalities, the result could be a severe limitation on their ability to serve the public.

B

The Court searches at length — and in vain — for legal authority to buttress its policy judgment. Despite its general statements to the contrary, the Court can find no support for its position in the debates on the civil rights legislation that included § 1983. Indeed, the legislative record suggests that *671the Members of the 42d Congress would have been dismayed by this ruling. ' Nor, despite its frequent citation of authorities that are only marginally relevant, can the Court rely on the traditional or current law of municipal tort liability. Both in the 19th century and now, courts and legislatures have recognized the importance of limiting the liability of local governments for official torts. Each of these conventional sources of law points to the need for qualified immunity for local governments. .

1

The modern dispute over municipal liability under § 1983 has focused on the defeat of the Sherman amendment during the deliberations on the Civil Rights Act of 1871. E. g., Monroe v. Pape, 365 U. S., at 187-191; Monell v. New York City Dept. of Social Services, 436 U. S., at 664-683. Senator Sherman proposed that local governments be held vicariously liable for constitutional deprivations caused by riots within their boundaries. As originally drafted, the measure imposed liability even if municipal officials had no actual knowledge of the impending disturbance.13 The amendment, which did not affect fche part of the Civil Rights Act that we know as § 1983, was approved by the Senate but rejected by the House of Representatives. 436 U. S., at 666. After two revisions by Conference Committees, bgth Houses passed what is now codified as 42 U. S. C. § 1986. The final version applied not just to local governments but to all “persons,” and it imposed no *672liability unless the defendant knew that a wrong was “about to be committed.”14

Because Senator Sherman initially proposed strict municipal liability for constitutional torts, the discussion of his amendment offers an invaluable insight into the attitudes of his colleagues on the question now before the Court. Much of the resistance to the measure flowed from doubts as to Congress’ power to impose vicarious liability on local governments. Monell v. New York City Dept. of Social Services, 436 U. S., at 673-683; id., at 706 (Powell, J., concurring). But opponents óf the amendment made additional arguments that strongly support recognition of qualified municipal immunity under § 1983.

First, several legislators expressed trepidation that the proposal’s strict liability approach could bankrupt local governments. They warned that liability under the proposal could bring municipalities “to a dead stop.” Cong. Globe, 42d Cong., 1st Sess., 763 (1871) (Sen. Casserly). See id., at 762 (Sen. Stevenson); id., at 772 (Sen. Thurman). Representative Bingham argued that municipal liability might be so great under the measure as to deprive a community “of the means of administering justice.” Id., at 798. Some Congressmen argued that strict liability would inhibit the effective operation of municipal corporations. The possibility of liability, Representative Kerr insisted, could prevent local officials from exercising “necessary and customary functions.” Id., at 789. See id., at 763 (Sen. Casserly); id., at 808 (Rep. Garfield).

*673Most significant, the opponents objected to liability imposed without any showing that a municipality knew of an impending constitutional deprivation. Senator Sherman defended this feature of the amendment as a characteristic of riot Acts long in force in England and this country. Id., at 760. But Senator Stevenson argued against creating “a corporate liability for personal injury which no prudence or foresight could have prevented.” Id., at 762. In the most thorough critique of the amendment, Senator Thurman carefully reviewed the riot Acts of Maryland and New York. He emphasized that those laws imposed liability only when a plaintiff proved that the local government had both notice of the impending injury and the power to prevent it. Id., at 771.

“Is not that right? Why make the county, or town, or parish liable when it had no reason whatsoever, to anticipate that any such crime was about to be committed, and when it had no knowledge of the commission of the crime until after it was committed? What justice is there in that?” Ibid.

These concerns were echoed in the House of Representatives. Representative Kerr complained that “it is not required, before liability shall attach, that it shall be known that there was any intention to commit these crimes, so as to fasten liability justly upon the municipality.” Id., at 788. He denounced the “total and absolute absence of notice, constructive or implied, within any decent limits of law or reason,” adding that the proposal “takes the property of one and gives it to another by mere force, without right, in the absence of guilt or knowledge, or the possibility of either.” Ibid. Similarly, Representative Willard argued that liability “is only warranted when the community . . . has proved faithless to its duties. . . .” Id., at 791. He criticized the absence of a requirement that it be “prov[ed] in court that there has been any default, any denial, any neglect on the part of *674the county, city, town, or parish to give citizens the full protection of the laws.” Ibid.

Partly in response to these objections, the amendment as finally enacted conditioned liability on a demonstration that the defendant knew that constitutional rights were about to be denied. Representative Poland introduced the new measure, noting that “any person who has knowledge of any of the offenses named . . . shall [have a] duty to use all reasonable diligence within his power to prevent it.” Id., at 804 (emphasis supplied). The same point was made by Representative Shellabarger, the sponsor of the entire Act and, with Representative Poland, a member of the Conference Committee that produced the final draft. Id., at 804-805; see id., at 807 (Rep: Garfield).

On the Senate side, one conferee stated that under the final version

“in order to make the [municipal] corporation liable as a body it must appear in some way to the satisfaction of the jury that the officers of the corporation, those persons whose duty it was to repress tumult, if they could, had reasonable notice of the fact that there was a tumult, or was likely to be one, and neglected to take the necessary means to prevent it.” Id., at 821 (Sen. Edmunds).

Senator Sherman disliked the revised provision. He complained that “before you can make [a person] responsible you have got to show that they had knowledge that the specific wrongs upon the particular person were about to be wrought.” Ibid.15

These objections to the Sherman amendment apply with equal force to strict municipal liability under § 1983. Just *675as the 42d Congress refused to hold municipalities vicariously liable for deprivations that could not be known beforehand, this Court should not hold those entities strictly liable for deprivations caused by actions that reasonably and in good faith were thought to be legal. The Court’s approach today, like the Sherman amendment, could spawn onerous judgments against local governments and distort the decisions of officers who fear municipal liability for their actions. Congress’ refusal to impose those burdens in 1871 surely undercuts any historical argument that federal judges should do so now.

The Court declares that its rejection of qualified immunity is “compelled” by the “legislative purpose” in enacting § 1983. Ante, at 650. One would expect powerful documentation to back up such a strong' statement. Yet the Court notes only three features of the legislative history of the Civil Rights Act. Far from “compelling” the Court’s strict liability approach, those features of the congressional record provide scant support for its position.

First, the Court reproduces statements by Congressmen attesting to the broad remedial scope of the law. Ante, at 636, and n. 17. In view of our many decisions recognizing the immunity of officers under § 1983, supra, at 666-667, those statements plainly shed no light on congressional intent with respect to immunity under the statute. Second, the Court cites Senator Stevenson’s remark that frequently “a statutory liability has been created against municipal corporations for injuries resulting from a neglect of corporate duty.” Ante, at 642-643, citing Cong. Globe, 42d Cong., 1st Sess., 762 (1871). The Senator merely stated the unobjectionable proposition that municipal immunity could be qualified or abolished by statute. This fragmentary observation provides no basis for the Court’s version of the legislative history.

Finally, the Court emphasizes the lack of comment on municipal immunity when opponents of the bill did discuss the immunities of government officers. “Had there been a *676similar common-law immunity for municipalities, the bill’s opponents doubtless would have raised the spectre of its destruction, as well.” Ante, at 643-644. This is but another example of the Court’s continuing willingness to find meaning in silence. This example is particularly noteworthy because the very next sentence in the Court’s opinion concedes: “To be sure, there were two doctrines that afforded municipal corporations some measure of protection from tort liability.” Ante, at 644. Since the opponents of the Sherman amendment repeatedly expressed their conviction that strict municipal liability was unprecedented and unwise, the failure to recite the theories of municipal immunity is of no relevance here. In any event, that silence cannot contradict the many contemporary judicial decisions applying that immunity. See infra, at 677-678, and nn. 16, 17.

2

The Court’s decision also runs counter to the common law in the 19th century, which recognized substantial tort immunity for municipal actions. E. g., 2 J. Dillon, Law of Municipal Corporations §§ 753, 764, pp. 862-863; 875-876 (2d ed. 1873); W. Williams, Liability of Municipal Corporations for Tort 9, 16 (1901). Nineteenth-century courts generally held that municipal corporations were not liable for acts undertaken in their “governmental,” as opposed to their “proprietary,” capacity.16 Most States now use other criteria *677for determining when a local government should be liable for damages. See infra, at 681-683. Still, the governmental/ proprietary distinction retains significance because it was so widely accepted when § 1983 was enacted. It is inconceivable that a Congress thoroughly versed in current legal doctrines, see Monell v. New York City Dept. of Social Services, 436 U. S., at 669, would have intended through silence to create the strict liability regime now imagined by this Court.

More directly relevant to this case is the common-law distinction between the “discretionary” and “ministerial” duties of local governments. This Court wrote in Harris v. District of Columbia, 256 U. S. 650, 652 (1921): “[W]hen acting in good faith municipal corporations are not liable for the manner in which they exercise discretionary powers of a public or legislative character.” See Weightmak v. The Corporation of Washington, 1 Black 39, 49-50 (1862). The rationale for this immunity derives from the theory of separation of powers. In Carr v. The Northern Liberties, 35 Pa. 324, 329 (1860), the Pennsylvania Supreme Court explained why a local government was immune from recovery for damage caused by an inadequate town drainage plan.

“[H]ow careful we must be that courts and juries do not encroach upon the functions committed to other public officers. It belongs to the province of town councils to direct the drainage of our towns, according to the best of their means and discretion, and we cannot directly or indirectly control them in either. No law allows us to substitute the judgment of a jury ... for that of the representatives of the town itself, to whom the business is especially committed by law.”

*678That reasoning, frequently applied in the 19th century,17 parallels the theory behind qualified immunity under § 1983. This Court has recognized the importance of preserving the autonomy of executive bodies entrusted with discretionary powers. Scheuer v. Rhodes held that executive officials who have broad responsibilities must enjoy a “range of discretion [that is] comparably broad.” 416 U. S., at 247. Consequently, the immunity available under § 1983 varies directly with “the scope of discretion and responsibilities of the office. . . .” 416 U. S., at 247. Strict municipal liability can only undermine that discretion.18

*679The lack of support for the Court’s view of the common law is evident in its reliance on Thayer v. Boston, 36 Mass. 511 (1837), as its principal authority. Ante, at 641-642. Thayer did hold broadly that a city could be liable for the authorized acts of its officers. 36 Mass., at 516. But Thayer was limited severely by later Massachusetts decisions. Bigelow v. Inhabitants of Randolph, 80 Mass. 541, 544-545 (1860), ruled that Thayer applied only to situations involving official malfeasance — or wrongful, bad-faith actions — not to actions based on neglect or nonfeasance. See Child v. Boston, 86 Mass. 41 (1862); Buttrick v. Lowell, 83 Mass. 172 (1861). Finally, Hill v. Boston, 122 Mass. 344, 359 (1877), squarely repudiated the broad holding of Thayer and limited municipal liability to acts performed in the proprietary interest of the municipality.19

*6803

Today’s decision also conflicts with the current law in 44 States and the District of Columbia. All of those jurisdictions provide municipal immunity at least analogous to a “good faith” defense against liability for constitutional torts. Thus, for municipalities in almost 90% of our jurisdictions, the Court creates broader liability for constitutional deprivations than for state-law torts.

*681Twelve States have laws creating municipal tort liability but barring damages for injuries caused by discretionary decisions or by the good-faith execution of a validly enacted, though unconstitutional, regulation.20 Municipalities in those States have precisely the form of qualified immunity that this Court has granted to executive officials under § 1983. Another 11 States provide even broader immunity for local governments. Five of those have retained the governmental/proprietary distinction,21 while Arkansas and South Dakota grant even broader protection for municipal corporations.22 Statutes in four more States protect local governments from tort liability except for particular injuries not relevant to this case, such as those due to motor vehicle accidents or negligent maintenance of public facilities.23 In *682Iowa, local governments are not liable for injuries caused by the execution with due care of any “officially enacted” statute or regulation.24

Sixteen States and the District of Columbia follow the traditional rule against recovery for damages imposed by discretionary decisions that are confided to particular officers or organs of government.25 Indeed, the leading commentators on governmental tort liability have noted both, the appropriateness and general acceptance of municipal immunity for discretionary acts. See Restatement (Second) of Torts § 895C (2) and Comment g (1979); K. Davis, Administrative Law of the Seventies § 25.13 (1976); W. Prosser, Law of Torts 986-987 (4th ed. 1971). In four States, local governments enjoy complete immunity from tort actions unless they have taken out liability insurance.26 Only five States *683impose the kind of blanket liability constructed by the Court today.27

C

The Court turns a blind eye to this overwhelming evidence that municipalities have enjoyed a qualified immunity and to the policy considerations that for the life of this Republic have justified its retention. This disregard of precedent and policy is especially unfortunate because suits under § 1983 typically implicate evolving constitutional standards. A good-faith defense is much more important for those actions than in those involving ordinary tort liability. The duty not to run over a pedestrian with a municipal bus is far less likely to change than is the rule as to what process, if any, is due the busdriver if he claims the right to a hearing after discharge.

The right of a discharged government employee to a “name clearing” hearing was not recognized until our decision in Board of Regents v. Roth, 408 U. S. 564 (1972). That ruling was handed down 10 weeks after Owen was discharged and 8 weeks after the city denied his request for a hearing. By stripping the city of any immunity, the Court punishes it for failing to predict our decision in Roth. As a result, local governments and their officials will face the unnerving prospect of crushing damages judgments whenever a policy valid under current law is later found to be unconstitutional. I can see no justice or wisdom in that outcome.

Justice Breyer

delivered the opinion of the Court.

In Monell v. New York City Dept. of Social Servs., 436 U. S. 658 (1978), this Court held that civil rights plaintiffs *31 suing a municipal entity under 42 U. S. C. § 1983 must show that their injury was caused by a municipal policy or custom. The case before the Court in Monell directly involved monetary damages. The question presented is whether the “policy or custom” requirement also applies when plaintiffs seek prospective relief, such as an injunction or a declaratory judgment. We conclude that it does so apply.

I

The case arises out of the following circumstances: The California Child Abuse and Neglect Reporting Act, Cal. Penal Code Ann. § 11164 et seq. (West Rev. Supp. 2010), requires law enforcement and other state agencies to investigate allegations of child abuse. These agencies must report to the California Department of Justice all instances of reported child abuse the agency finds “not unfounded,” even if they are “inconclusive or unsubstantiated.” §§ 11169(a), 11170(a)(3). The statute requires the department to include all these reports in a Child Abuse Central Index (Index), where they remain available to various state agencies for at least 10 years. § 11170(a). The statute also says that if

“a report has previously been filed which subsequently proves to be unfounded, the Department of Justice shall be notified in writing of that fact and shall not retain the report.” § 11169(a).

The statute, however, does not set forth procedures for reviewing whether a previously filed report is unfounded, or for allowing individuals to challenge their inclusion in the Index. Nor, up until the time of this lawsuit, had California or Los Angeles County created any such procedures. But cf. § 11170(a)(2) (“The submitting agencies are responsible for the accuracy, completeness, and retention of the reports described in this section”).

The two plaintiffs in this case were initially accused of child abuse. But they were later exonerated. They sought to have their names removed from the Index. Unable to *32 convince the Los Angeles Sheriff’s Department to remove them, they filed this § 1983 case against the attorney general of California, the Los Angeles County sheriff, two detectives in the sheriff’s department, and the County of Los Angeles. They sought damages, an injunction, and a declaration that the defendants had deprived them of their constitutional rights by failing to create a procedural mechanism through which one could contest inclusion on the Index. See U. S. Const., Arndt. 14; Rev. Stat. § 1979, 42 U. S. C. § 1983. The District Court for the Central District of California granted summary judgment to all of the defendants on the ground that California had not deprived the plaintiffs of a constitutionally protected “liberty” interest. But on appeal the Ninth Circuit disagreed.

The Ninth Circuit held that the Fourteenth Amendment required the State to provide those included on the list notice and “‘some kind of hearing.’” 554 F. 3d 1170, 1201 (2009). Thus the Circuit held that the plaintiffs were entitled to declaratory relief, and it believed that (on remand) they might prove damages as well. Ibid.

The Ninth Circuit also held that the plaintiffs were prevailing parties, thereby entitled to approximately $600,000 in attorney’s fees. 42 U. S. C. § 1988(b) (providing for payment of attorney’s fees to parties prevailing on §1983 claims). See No. 05-56467 (June 22, 2009), App. to Pet. for Cert. 1-4 (hereinafter First Fee Order); No. 05-56467 (Dec. 2, 2009), App. to Reply to Brief in Opposition 1-2 (hereinafter Second Fee Order). The Ninth Circuit wrote that Los Angeles County must pay approximately $60,000 of this amount. First Fee Order 3; Second Fee Order 2.

Los Angeles County denied that it was liable and therefore that it could be held responsible for attorney’s fees. It argued that, in respect to the county, the plaintiffs were not prevailing parties. That is because the county is a municipal entity. Under Monell’s holding a municipal entity is liable under § 1983 only if a municipal “policy or custom” caused a *33 plaintiff to be deprived of a federal right. 436 U. S., at 694 (emphasis added). And it was state policy, not county policy, that brought about any deprivation here.

The Ninth Circuit responded to this argument as follows: First, it said that county policy might be responsible for the deprivation. It “is possible,” the Ninth Circuit said, that the county, “[b]y failing to” “creat[e] an independent procedure that would allow” the plaintiffs “to challenge their listing^] . . . adopted a custom and policy that violated” the plaintiffs’ “constitutional rights.” 554 F. 3d, at 1202. Second, it said that “because this issue is not clear based on the record before us on appeal ... we remand to the district court to determine the County’s liability under Monell.” Ibid. Third, it saw no reason to remand in respect to the county’s obligation to pay $60,000 in attorney’s fees. That, it wrote, is because “in our circuit ... the limitations to liability established in Monell do not apply to claims for prospective relief,” such as the declaratory judgment that the Circuit had ordered entered. First Fee Order 3-4 (citing Chaloux v. Killeen, 886 F. 2d 247, 250 (CA9 1989); Truth v. Kent School Dist., 542 F. 3d 634, 644 (CA9 2008); emphasis added).

The county then asked us to review this last-mentioned Ninth Circuit holding, namely, the holding that Monell’s “policy or custom” requirement applies only to claims for damages but not to claims for prospective relief. Because the Courts of Appeals are divided on this question, we granted the county’s petition for certiorari. Compare Reynolds v. Giuliani, 506 F. 3d 183, 191 (CA2 2007) (holding that Monell’s “policy or custom” requirement applies to claims for prospective relief as well as claims for damages); Dirrane v. Brookline Police Dept., 315 F. 3d 65, 71 (CA1 2002) (same); Greensboro Professional Fire Fighters Assn., Local 3157 v. Greensboro, 64 F. 3d 962, 967, n. 6 (CA4 1995) (applying the Monell requirement to a prospective relief claim); Church v. Huntsville, 30 F. 3d 1332, 1347 (CA11 1994) (same), with *34 Chaloux, supra, at 251 (holding that Monell does not apply to prospective relief claims). See also Gernetzke v. Kenosha Unified School Dist. No. 1, 274 F. 3d 464, 468 (CA7 2001) (reserving the question but noting the “predominant” view that “Monell’s holding applies regardless of the nature of the relief sought”).

We conclude that Monell’s holding applies to § 1983 claims against municipalities for prospective relief as well as to claims for damages.

II

A

We begin with § 1983 itself, which provides:

“Every person who, under color of any [state] statute, ordinance, regulation, custom, or usage . . . subjects, or causes to be subjected, any . .. other person ... to the deprivation of any rights ... secured by the Constitution and laws [of the United States], shall be liable to the party injured in an action at law, suit in equity, or other proper proceeding for redress.” (Emphasis added.)

In 1961, in Monroe v. Pape, 365 U. S. 167, this Court held that municipal entities were not “person[s]” under §1983. The Court based this conclusion on the history of the Civil Rights Act of 1871’s enactment. It noted that Congress rejected an amendment (called the Sherman amendment) that would have made municipalities liable for damage done by private persons “‘riotously and tumultuously assembled.’” Id., at 188-190, and n. 38 (quoting Cong. Globe, 42d Cong., 1st Sess., 663 (1871)). This rejection, the Court thought, reflected a determination by the 1871 House of Representatives that “ ‘Congress had no constitutional power to impose any obligation upon county and town organizations, the mere instrumentality for the administration of state law.’” 365 U. S., at 190 (quoting Cong. Globe, supra, at 804 (statement of Rep. Poland); emphasis added). The Court concluded that Congress must have doubted its “constitutional *35 power ... to impose civil liability on municipalities.” 365 U. S., at 190. And for that reason, Congress must have intended to exclude municipal corporations as §1983 defendants. The statute’s key term “person” therefore did not cover municipal entities. Id., at 191.

Sixteen years later, in Monell, the Court reconsidered the question of municipal liability. After reexamining the 1871 legislative history in detail, the Court concluded that Congress had rejected the Sherman amendment, not because it would have imposed liability upon municipalities, but because it would have imposed liability upon municipalities based purely upon the acts of others. That is to say, the rejected amendment would have imposed liability upon local governments “without regard to whether a local government was in any way at fault for the breach of the peace for which it was to be held for damages.” 436 U. S., at 681, n. 40 (emphasis added). In Monell’s view Congress may have thought that it lacked the power to impose that kind of indirect liability upon municipalities, id., at 679, but “nothing said in debate on the Sherman amendment would have prevented holding a municipality liable... for its own violations of the Fourteenth Amendment,” id., at 683 (emphasis added). The Court, overruling Monroe, held that municipalities were “persons” under § 1983. 436 U. S., at 690.

The Court also concluded that a municipality could not be held liable under § 1983 solely because it employed a tort-feasor. The Court’s conclusion rested on “the language of § 1983, read against the background of the same legislative history.” Id., at 691. Section 1983’s causation language imposes liability on a “ ‘person who . . . shall subject, or cause to be subjected, any person’” to a deprivation of federal rights. Ibid. (quoting 17 Stat. 13; emphasis deleted). That language, the Court observed, could not “be easily read to impose liability vicariously . . . solely on the basis of the existence of an employer-employee relationship with a tort-feasor.” 436 U. S., at 692. The statute’s legislative history, *36 in particular the constitutional objections that had been raised to the Sherman amendment, supported this conclusion. Id., at 692-694, and n. 57.

For these reasons, the Court concluded that a municipality could be held liable under § 1983 only for its own violations of federal law. Id., at 694. The Court described what made a violation a municipality’s own violation:

“Local governing bodies, therefore, can be sued directly under § 1983 for monetary, declaratory, or injunctive relief where, as here, the action that is alleged to be unconstitutional implements or executes a policy statement, ordinance, regulation, or decision officially adopted and promulgated by that body’s officers. . . . [They can also be sued for] deprivations visited pursuant to governmental ‘custom’ even though such a custom has not received formal approval through the body’s official decisionmaking channels.” Id., at 690-691 (footnote omitted).

The Court has also included the terms “usage” and “practice” as customs for which liability is appropriate. See ibid. The length of this list of types of municipal action leads us here to use a shorthand term “policy or custom,” but when we do so, we mean to refer to the entire list. See id., at 694 (using the shorthand “policy or custom”); see also, e. g., Fitzgerald v. Barnstable School Comm., 555 U. S. 246, 257-258 (2009) (using the phrase “custom, policy, or practice,” to describe municipal liability under § 1983).

In sum, in Monell the Court held that “a municipality cannot be held liable” solely for the acts of others, e. g., “solely because it employs a tortfeasor.” 436 U. S., at 691. But the municipality may be held liable “when execution of a government’s policy or custom . . . inflicts the injury.” Id., at 694 (emphasis added).

B

The language of §1983 read in light of Monell’s understanding of the legislative history explains why claims for *37 prospective relief, like claims for money damages, fall within the scope of the “policy or custom” requirement. Nothing in the text of § 1983 suggests that the causation requirement contained in the statute should change with the form of relief sought. In fact, the text suggests the opposite when it provides that a person who meets § 1983’s elements “shall be liable ... in an action at law, suit in equity, or other proper proceeding for redress.” Thus, as Monell explicitly stated, “[l]ocal governing bodies . . . can be sued directly under § 1983 for monetary, declaratory, or injunctive relief where, as here, the action that is alleged to be unconstitutional implements or executes” a policy or custom. 436 U. S., at 690 (emphasis added). Monell went on to quote this Court’s statement in a 1973 case, Kenosha v. Bruno, 412 U. S. 507, 513, to the effect that the Congress that enacted § 1983 did not intend the “ ‘generic word “person” ... to have a bifurcated application to municipal corporations depending on the nature of the relief sought against them.’” 436 U. S., at 701, n. 66 (emphasis added). Monell added that “[n]othing we say today affects” this pre-Monell “conclusion.” Ibid.

Monell’s logic also argues against any such relief-based bifurcation. The Monell Court thought that Congress intended potential § 1983 liability where a municipality’s own violations were at issue but not where only the violations of others were at issue. The “policy or custom” requirement rests upon that distinction and embodies it in law. To find the requirement inapplicable where prospective relief is at issue would undermine Monell’s logic. For whether an action or omission is a municipality’s “own” has to do with the nature of the action or omission, not with the nature of the relief that is later sought in court.

C

The Humphrieses’ (hereinafter respondents) arguments to the contrary are unconvincing. Respondents correctly note that by the time Monell reached the Supreme Court only the plaintiffs’ damages claim remained live. See id., at 661. *38 From this fact they conclude that the Court’s holding applies directly only to claims for monetary damages. A holding, however, can extend through its logic beyond the specific facts of the particular case. It does so here.

Respondents add that not only did Monell involve a damages claim, but its holding rests upon the concern that municipalities might have to pay large damages awards. The Court so suggests when it points out that municipalities should not be liable for an employee’s wrongful acts, simply by applying agency-based principles of respondeat superior. But as we have pointed out, the Court’s rejection of respondeat superior liability primarily rested not on the municipality’s economic needs, but on the fact that liability in such a case does not arise out of the municipality’s own wrongful conduct.

Respondents further claim that, where prospective relief is at issue, Monell is redundant. They say that a court cannot grant prospective relief against a municipality unless the municipality’s own conduct has caused the violation. Hence, where such relief is otherwise proper, the Monell requirement “shouldn’t screen out any case.” Tr. of Oral Arg. 48.

To argue that a requirement is necessarily satisfied, however, is not to argue that its satisfaction is unnecessary. If respondents are right, our holding may have limited practical significance. But that possibility does not provide us with a convincing reason to sow confusion by adopting a bifurcated relief-based approach to municipal liability that the Court has previously rejected.

Finally, respondents make the mirror-image argument that applying Monell's requirement to prospective relief claims will leave some set of ongoing constitutional violations beyond redress. Despite the fact that four Circuits apply Monell's requirement to prospective relief, however, respondents have not presented us with any actual or hypothetical example that provides serious cause for concern.

*39 * h= *

For these reasons, we hold that MonelVs “policy or custom” requirement applies in §1983 cases irrespective of whether the relief sought is monetary or prospective. The Ninth Circuit’s contrary judgment is reversed, and the ease is remanded for farther proceedings consistent with this opinion.

It is so ordered.

Justice Brennan

delivered the opinion of the Court, except as to Part II-B.

In Monell v. New York City Dept. of Social Services, 436 U. S. 658 (1978), the Court concluded that municipal liability-under 42 U. S. C. § 1983 is limited to deprivations of federally protected rights caused by action taken “pursuant to official municipal policy of some nature . . . .” Id., at 691. The question presented is whether, and in what circumstances, a decision by municipal policymakers on a single occasion may satisfy this requirement.

I

Bertold Pembaur is a licensed Ohio physician and the sole proprietor of the Rockdale Medical Center, located in the city of Cincinnati in Hamilton County. Most of Pembaur’s patients are welfare recipients who rely on government assistance to pay for medical care. During the spring of 1977, Simon Leis, the Hamilton County Prosecutor, began investigating charges that Pembaur fraudulently had accepted payments from state welfare agencies for services not actually provided to patients. A grand jury was convened, and the case was assigned to Assistant Prosecutor William Whalen. *472In April, the grand jury charged Pembaur in a six-count indictment.

During the investigation, the grand jury issued subpoenas for the appearance of two of Pembaur’s employees. When these employees failed to appear as directed, the Prosecutor obtained capiases for their arrest and detention from the Court of Common Pleas of Hamilton County.1

On May 19,1977, two Hamilton County Deputy Sheriffs attempted to serve the capiases at Pembaur’s clinic. Although the reception area is open to the public, the rest of the clinic may be entered only through a door next to the receptionist’s window. Upon arriving, the Deputy Sheriffs identified themselves to the receptionist and sought to pass through this door, which was apparently open. The receptionist blocked their way and asked them to wait for the doctor. When Pembaur appeared a moment later, he and the receptionist closed the door, which automatically locked from the inside, and wedged a piece of wood between it and the wall. Returning to the receptionist’s window, the Deputy Sheriffs identified themselves to Pembaur, showed him the capiases and explained why they were there. Pembaur refused to let them enter, claiming that the police had no legal authority to be there and requesting that they leave. He told them that he had called the Cincinnati police, the local media, and his lawyer. The Deputy Sheriffs decided not to take further action until the Cincinnati police arrived.

Shortly thereafter, several Cincinnati police officers appeared. The Deputy Sheriffs explained the situation to them and asked that they speak to Pembaur. The Cincinnati police told Pembaur that the papers were lawful and that he should allow the Deputy Sheriffs to enter. When Pembaur refused, the Cincinnati police called for a superior officer. When he too failed to persuade Pembaur to open the door, *473the Deputy Sheriffs decided to call their supervisor for further instructions. Their supervisor told them to call Assistant Prosecutor Whalen and to follow his instructions. The Deputy Sheriffs then telephoned Whalen and informed him of the situation. Whalen conferred with County Prosecutor Leis, who told Whalen to instruct the Deputy Sheriffs to "go in and get [the witnesses].” Whalen in turn passed these instructions along to the Deputy Sheriffs.

After a final attempt to persuade Pembaur voluntarily to allow them to enter, the Deputy Sheriffs tried unsuccessfully to force the door. City police officers, who had been advised of the County Prosecutor’s instructions to “go in and get” the witnesses, obtained an axe and chopped down the door. The Deputy Sheriffs then entered and searched the clinic. Two individuals who fit descriptions of the witnesses sought were detained, but turned out not to be the right persons.

After this incident, the Prosecutor obtained an additional indictment against Pembaur for obstructing police in the performance of an authorized act. Although acquitted of all other charges, Pembaur was convicted for this offense. The Ohio Court of Appeals reversed, reasoning that Pembaur was privileged under state law to exclude the deputies because the search of his office violated the Fourth Amendment. State v. Pembaur, No. C-790380 (Hamilton County Court of Appeals, Nov. 3, 1982). The Ohio Supreme Court reversed and reinstated the conviction. State v. Pembaur, 9 Ohio St. 3d 136, 459 N. E. 2d 217, cert. denied, 467 U. S. 1219 (1984). The Supreme Court held that the state-law privilege applied only to bad-faith conduct by law enforcement officials, and that, under the circumstances of this case, Pembaur was obliged to acquiesce to the search and seek redress later in a civil action for damages. 9 Ohio St. 3d, at 138, 459 N. E. 2d, at 219.

On April 20, 1981, Pembaur filed the present action in the United States District Court for the Southern District of Ohio against the city of Cincinnati, the County of Hamilton, *474the Cincinnati Police Chief, the Hamilton County Sheriff, the members of the Hamilton Board of County Commissioners (in their official capacities only), Assistant Prosecutor Whalen, and nine city and county police officers.2 Pembaur sought damages under 42 U. S. C. § 1983, alleging that the county and city police had violated his rights under the Fourth and Fourteenth Amendments. His theory was that, absent exigent circumstances, the Fourth Amendment prohibits police from searching an individual’s home or business without a search warrant even to execute an arrest warrant for a third person. We agreed with that proposition in Steagald v. United States, 451 U. S. 204 (1981), decided the day after Pembaur filed this lawsuit. Pembaur sought $10 million in actual and $10 million in punitive damages, plus costs and attorney’s fees.

Much of the testimony at the 4-day trial concerned the practices of the Hamilton County Police in serving capiases. Frank Webb, one of the Deputy Sheriffs present at the clinic on May 19, testified that he had previously served capiases on the property of third persons without a search warrant, but had never been required to use force to gain access. Assistant Prosecutor Whalen was also unaware of a prior instance in which police had been denied access to a third person’s property in serving a capias and had used force to gain entry. Lincoln Stokes, the County Sheriff, testified that the Department had no written policy respecting the serving of capiases on the property of third persons and that the proper response in any given situation would depend upon the circumstances. He too could not recall a specific instance in *475which entrance had been denied and forcibly gained. Sheriff Stokes did testify, however, that it was the practice in his Department to refer questions to the County Prosecutor for instructions under appropriate circumstances and that “it was the proper thing to do” in this case.

The District Court awarded judgment to the defendants and dismissed the complaint in its entirety. The court agreed that the entry and search of Pembaur’s clinic violated the Fourth Amendment under Steagald, supra, but held Steagald inapplicable since it was decided nearly four years after the incident occurred. Because it construed the law in the Sixth Circuit in 1977 to permit law enforcement officials to enter the premises of a third person to serve a capias, the District Court held that the individual municipal officials were all immune under Harlow v. Fitzgerald, 457 U. S. 800 (1982).

The claims against the county and the city were dismissed on the ground that the individual officers were not acting pursuant to the kind of “official policy” that is the predicate for municipal liability under Monell. With respect to Hamilton County, the court explained that, even assuming that the entry and search were pursuant to a governmental policy, “it was not a policy of Hamilton County per se” because “[t]he Hamilton County Board of County Commissioners, acting on behalf of the county, simply does not establish or control the policies of the Hamilton County Sheriff.” With respect to the city of Cincinnati, the court found that “the only policy or custom followed . . . was that of aiding County Sheriff’s Deputies in the performance of their duties.” The court found that any participation by city police in the entry and search of The clinic resulted from decisions by individual officers as to the permissible scope of assistance they could provide, and not from a city policy to provide this particular kind of assistance.

On appeal, Pembaur challenged only the dismissal of his claims against Whalen, Hamilton County, and the city of Cin*476cinnati. The Court of Appeals for the Sixth Circuit upheld the dismissal of Pembaur’s claims against Whalen and Hamilton County, but reversed the dismissal of his claim against the city of Cincinnati on the ground that the District Court’s findings concerning the policies followed by the Cincinnati police were clearly erroneous. 746 F. 2d 337 (1984).3

The Court of Appeals affirmed the District Court’s dismissal of Pembaur’s claim against Hamilton County, but on different grounds. The court held that the County Board’s lack of control over the Sheriff would not preclude county liability if “the nature and duties of the Sheriff are such that his acts may fairly be said to represent the county’s official policy with respect to the specific subject matter.” Id., at 340-341. Based upon its examination of Ohio law, the Court of Appeals found it “clea[r]” that the Sheriff and the Prosecutor were both county officials authorized to establish “the official policy of Hamilton County” with respect to matters of law enforcement. Id., at 341. Notwithstanding these conclusions, however, the court found that Pembaur’s claim against the county had been properly dismissed:

“We believe that Pembaur failed to prove the existence of a county policy in this case. Pembaur claims that the deputy sheriffs acted pursuant to the policies of the Sheriff and Prosecutor by forcing entry into the medical center. Pembaur has failed to establish, however, anything more than that, on this one occasion, the Prosecutor and the Sheriff decided to force entry into his office. . . . That single, discrete decision is insufficient, *477by itself, to establish that the Prosecutor, the Sheriff, or both were implementing a governmental policy.” Ibid. (footnote omitted) (emphasis in original).

Pembaur petitioned for certiorari to review only the dismissal of his claim against Hamilton County. The decision of the Court of Appeals conflicts with holdings in several other Courts of Appeals,4 and we granted the petition to resolve the conflict. 472 U. S. 1016 (1985). We reverse.

h-i t — I

A

Our analysis must begin with the proposition that “Congress did not intend municipalities to be held liable unless action pursuant to official municipal policy of some nature caused a constitutional tort.” Monell v. New York City Dept. of Social Services, 436 U. S., at 691.5 As we read its opinion, the Court of Appeals held that a single decision to *478take particular action, although made by municipal policymakers, cannot establish the kind of “official policy” required by Monell as a predicate to municipal liability under § 1983.6 The Court of Appeals reached this conclusion without referring to Monell — indeed, without any explanation at all. However, examination of the opinion in Monell clearly demonstrates that the Court of Appeals misinterpreted its holding.

Monell is a case about responsibility. In the first part of the opinion, we held that local government units could be made liable under § 1983 for deprivations of federal rights, overruling a contrary holding in Monroe v. Pape, 365 U. S. 167 (1961). In the second part of the opinion, we recognized a limitation on this liability and concluded that a municipality cannot be made liable by application of the doctrine of respondeat superior. See Monell, 436 U. S., at 691. In part, this conclusion rested upon the language of §1983, which imposes liability only on a person who “subjects, or causes to be subjected,” any individual to a deprivation of federal rights; we noted that this language “cannot easily be read to impose liability vicariously on government bodies solely on the basis of the existence of an employer-employee relationship -with a tortfeasor.” Id., at 692. Primarily, *479however, our conclusion rested upon the legislative history, which disclosed that, while Congress never questioned its power to impose civil liability on municipalities for their own illegal acts, Congress did doubt its constitutional power to impose such liability in order to oblige municipalities to control the conduct of others. Id., at 665-683.7 We found that, because of these doubts, Congress chose not to create such obligations in § 1983. Recognizing that this would be the effect of a federal law of respondeat superior, we concluded that § 1983 could not be interpreted to incorporate doctrines of vicarious liability. Id., at 692-694, and n. 57.

The conclusion that tortious conduct, to be the basis for municipal liability under §1983, must be pursuant to a municipality’s “official policy” is contained in this discussion. The “official policy” requirement was intended to distinguish acts of the municipality from acts of employees of the municipality, and thereby make clear that municipal liability is limited to action for which the municipality is actually responsi*480ble.8 Monell reasoned that recovery from a municipality is limited to acts that are, properly speaking, acts “of the municipality” — that is, acts which the municipality has officially sanctioned or ordered.

With this understanding, it is plain that municipal liability may be imposed for a single decision by municipal policymakers under appropriate circumstances. No one has ever doubted, for instance, that a municipality may be liable under § 1983 for a single decision by its properly constituted legislative body — whether or not that body had taken similar action in the past or intended to do so in the future — because even a single decision by such a body unquestionably constitutes an act of official government policy. See, e. g., Owen v. City of Independence, 445 U. S. 622 (1980) (City Council passed resolution firing plaintiff without a pretermination hearing); Newport v. Fact Concerts, Inc., 453 U. S. 247 (1981) (City Council canceled license permitting concert because of dispute over content of performance). But the power to establish policy is no more the exclusive province of the legislature at the local level than at the state or national level. Monell’s language makes clear that it expressly envisioned other officials “whose acts or edicts may fairly be said to represent official policy,” Monell, supra, at 694, and whose decisions therefore may give rise to municipal liability under § 1983.

Indeed, any other conclusion would be inconsistent with the principles underlying § 1983. To be sure, “official policy” often refers to formal rules or understandings — often but not always committed to writing — that are intended to, and do, establish fixed plans of action to be followed under similar cir*481cumstances consistently and over time. That was the case in Monell itself, which involved a written rule requiring pregnant employees to take unpaid leaves of absence before such leaves were medically necessary. However, as in Owen and Newport, a government frequently chooses a course of action tailored to a particular situation and not intended to control decisions in later situations. If the decision to adopt that particular course of action is properly made by that government’s authorized decisionmakers, it surely represents an act of official government “policy” as that term is commonly understood.9 More importantly, where action is directed by those who establish governmental policy, the municipality is equally responsible whether that action is to be taken only once or to be taken repeatedly. To deny compensation to the victim would therefore be contrary to the fundamental purpose of § 1983.

B

Having said this much, we hasten to emphasize that not every decision by municipal officers automatically subjects the municipality to §1983 liability. Municipal liability attaches only where the decisionmaker possesses final authority to establish municipal policy with respect to the action ordered.10 The fact that a particular official — even a policy-*482making official — has discretion in the exercise of particular functions does not, without more, give rise to municipal liability based on an exercise of that discretion. See, e. g., Oklahoma City v. Tuttle, 471 U. S., at 822-824.11 The offi*483cial must also be responsible for establishing final government policy respecting such activity before the municipality can be held liable.12 Authority to make municipal policy may be granted directly by a legislative enactment or may be delegated by an official who possesses such authority, and of course, whether an official had final policymaking authority is a question of state law. However, like other governmental entities, municipalities often spread policymaking authority among various officers and official bodies. As a result, particular officers may have authority to establish binding county policy respecting particular matters and to adjust that policy for the county in changing circumstances. To hold a municipality liable for actions ordered by such officers exercising their policymaking authority is no more an application of the theory of respondeat superior than was holding the municipalities liable for the decisions of the City Councils in Owen and Newport. In each case municipal liability attached to a single decision to take unlawful action made by municipal policymakers. We hold that municipal liability under §1983 attaches where — and only where — a deliberate choice to follow a course of action is made from among various alternatives by the official or officials responsible for establishing final policy with respect to the subject matter in ques*484tion. See Tuttle, supra, at 823 (“‘policy’ generally implies a course of action consciously chosen from among various alternatives”).

C

Applying this standard to the case before us, we have little difficulty concluding that the Court of Appeals erred in dismissing petitioner’s claim against the county. The Deputy Sheriffs who attempted to serve the capiases at petitioner’s clinic found themselves in a difficult situation. Unsure of the proper course of action to follow, they sought instructions from their supervisors. The instructions they received were to follow the orders of the County Prosecutor. The Prosecutor made a considered decision based on his understanding of the law and commanded the officers forcibly to enter petitioner’s clinic. That decision directly caused the violation of petitioner’s Fourth Amendment rights.

Respondent argues that the County Prosecutor lacked authority to establish municipal policy respecting law enforcement practices because only the County Sheriff may establish policy respecting such practices. Respondent suggests that the County Prosecutor was merely rendering “legal advice” when he ordered the Deputy Sheriffs to “go in and get” the witnesses. Consequently, the argument concludes, the action of the individual Deputy Sheriffs in following this advice and forcibly entering petitioner’s clinic was not pursuant to a properly established municipal policy.

We might be inclined to agree with respondent if we thought that the Prosecutor had only rendered “legal advice.” However, the Court of Appeals concluded, based upon its examination of Ohio law, that both the County Sheriff and the County Prosecutor could establish county policy under appropriate circumstances, a conclusion that we do not question here.13 Ohio Rev. Code Ann. § 309.09(A) (1979) *485provides that county officers may “require . . . instructions from [the County Prosecutor] in matters connected with their official duties.” Pursuant to standard office procedure, the Sheriff’s Office referred this matter to the Prosecutor and then followed his instructions. The Sheriff testified that his Department followed this practice under appropriate circumstances and that it was “the proper thing to do” in this case. We decline to accept respondent’s invitation to overlook this delegation of authority by disingenuously labeling the Prosecutor’s clear command mere “legal advice.” In ordering the Deputy Sheriffs to enter petitioner’s clinic the County Prosecutor was acting as the final decisionmaker for the county, and the county may therefore be held liable under § 1983.

The decision of the Court of Appeals is reversed, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

Justice White,

concurring.

The forcible entry made in this case was not then illegal under federal, state, or local law. The city of Cincinnati frankly conceded that forcible entry of third-party property to effect otherwise valid arrests was standard operating procedure. There is no reason to believe that respondent county would abjure using lawful means to execute the capiases issued in this case or had limited the authority of its officers to use force in executing capiases. Further, the county officials who had the authority to approve or disapprove such entries opted for the forceful entry, a choice that was later held to be inconsistent with the Fourth Amendment. Vesting discretion in its officers to use force and its use in this case sufficiently manifested county policy to warrant reversal of the judgment below.

*486This does not mean that every act of municipal officers with final authority to effect or authorize arrests and searches represents the policy of the municipality. It would be different if Steagald v. United States, 451 U. S. 204 (1981), had been decided when the events at issue here occurred, if the State Constitution or statutes had forbidden forceful entries without a warrant, or if there had been a municipal ordinance to this effect. Local law enforcement officers are expected to obey the law and ordinarily swear to do so when they take office. Where the controlling law places limits on their authority, they cannot be said to have the authority to make contrary policy. Had the Sheriff or Prosecutor in this case failed to follow an existing warrant requirement, it would be absurd to say that he was nevertheless executing county policy in authorizing the forceful entry in this case and even stranger to say that the county would be liable if the Sheriff had secured a warrant and it turned out that he and the Magistrate had mistakenly thought there was probable cause for the warrant. If deliberate or mistaken acts like this, admittedly contrary to local law, expose the county to liability, it must be on the basis of respondeat superior and not because the officers’ acts represent local policy.

Such results would not conform to Monell and the cases following it. I do not understand the Court to hold otherwise in stating that municipal liability attaches where “a deliberate choice to follow a course of action is made from among various alternatives by the official or officials responsible for establishing final policy with respect to the subject matter in question.” Ante, at 483-484. A sheriff, for example, is not the final policymaker with respect to the probable-cause requirement for a valid arrest. He has no alternative but to act in accordance with the established standard; and his deliberate or mistaken departure from the controlling law of arrest would not represent municipal policy.

In this case, however, the Sheriff and the Prosecutor chose a course that was not forbidden by any applicable law, a *487choice that they then had the authority to make. This was county policy, and it was no less so at the time because a later decision of this Court declared unwarranted forceful entry into third-party premises to be violation of the Fourth Amendment.* Hence, I join the Court’s opinion and judgment.

Justice Stevens,

concurring in part and concurring in the judgment.

This is not a hard case. If there is any difficulty, it arises from the problem of obtaining a consensus on the meaning of the word “policy” — a word that does not appear in the text of 42 U. S. C. § 1983, the statutory provision that we are supposed to be construing. The difficulty is thus a consequence of this Court’s lawmaking efforts rather than the work of the Congress of the United States.1

With respect to both the merits of the constitutional claim and the county’s liability for the unconstitutional activities of its agents performed in the course of their official duties, there can be no doubt that the Congress that enacted the Ku Klux Act in 1871 intended the statute to authorize a recovery in a case of this kind. When police officers chopped down the door to petitioner’s premises in order to serve capiases on two witnesses, they violated petitioner’s constitutional rights. Steagald v. United States, 451 U. S. 204 (1981), *488makes it perfectly clear that forcible entry to a third party’s premises to execute an arrest warrant is unconstitutional if the entry is without a search warrant and in the absence of consent or exigent circumstances.2 In my view, it is not at all surprising that respondents have “conceded” the retro-activity of Steagald. For Steagald plainly presented its holding as compelled by, and presaged in, well-established precedent.3

*489Similarly, if we view the question of municipal liability from the perspective of the Legislature that enacted the Ku Klux Act of 1871, the answer is clear. The legislative history indicating that Congress did not intend to impose civil liability on municipalities for the conduct of third parties, ante, at 478-479, and n. 7, merely confirms the view that it did intend to impose liability for the governments’ own illegal acts — including those acts performed by their agents in the course of their employment. In other words, as I explained in my dissent in Oklahoma City v. Tuttle, 471 U. S. 808, 835-840 (1985), both the broad remedial purpose of the statute and the fact that it embodied contemporaneous common-law doctrine, including respondeat superior, require a conclusion that Congress intended that a governmental entity be hable for the constitutional deprivations committed by its agents in the course of their duties.4

*490Finally, in construing the scope of § 1983, the Court has sometimes referred to “considerations of public policy.”5 To the extent that such “policy” concerns are relevant, they also support a finding of county liability. A contrary construction would produce a most anomalous result. The primary responsibility for protecting the constitutional rights of the residents of Hamilton County from the officers of Hamilton County should rest on the shoulders of the county itself, rather than on the several agents who were trying to perform their jobs. Although I recognize that the county may provide insurance protection for its agents, I believe that the primary party against whom the judgment should run is the county itself. The county has the resources and the authority that can best avoid future constitutional violations and provide a fair remedy for those that have occurred in the past. Thus, even if “public policy” concerns should inform the construction of § 1983, those considerations, like the statute’s remedial purpose and common-law background, support *491a conclusion of county liability for the unconstitutional, axe-swinging entry in this case.

Because I believe that Parts I, II-A, and II-C are consistent with the purpose and policy of § 1983, as well as with our precedents, I join those Parts of the Court’s opinion6 and concur in the judgment.

Justice O’Connor,

concurring in part and concurring in the judgment.

For the reasons stated by Justice White, I agree that the municipal officers here were acting as policymakers within the meaning of Monell v. New York City Dept. of Social Services, 436 U. S. 658 (1978). As the city of Cincinnati freely conceded, forcible entry of third-party property to effect an arrest was standard operating procedure in May 1977. Given that this procedure was consistent with federal, state, and local law at the time the case arose, it seems fair to infer that respondent county’s policy was no different. Moreover, under state law as definitively construed by the Court of Appeals, the county officials who opted for the forcible entry “had the authority to approve or disapprove such entries.” Ante, at 485 (White J., concurring). Given this combination of circumstances, I agree with Justice White that the decision to break down the door “sufficiently manifested county policy to warrant reversal of the judgment below.” Ibid. Because, however, I believe that the reasoning of the majority goes beyond that necessary to decide the case, and because I fear that the standard the majority articulates may be misread to expose municipalities to liability beyond that envisioned by the Court in Monell, I join only Parts I and II-A of the Court’s opinion and the judgment.

*492Justice Powell,

with whom The Chief Justice and Justice Rehnquist join, dissenting.

The Court today holds Hamilton County liable for the forcible entry in May 1977 by Deputy Sheriffs into petitioner’s office. The entry and subsequent search were pursuant to capiases for third parties — petitioner’s employees — who had failed to answer a summons to appear as witnesses before a grand jury investigating petitioner. When petitioner refused to allow the Sheriffs to enter, one of them, at the request of his supervisor, called the office of the County Prosecutor for instructions. The Assistant County Prosecutor received the call, and apparently was in doubt as to what advice to give. He referred the question to the County Prosecutor, who advised the Deputy Sheriffs to “go in and get them [the witnesses]” pursuant to the capiases. This five-word response to a single question over the phone is now found by this Court to have created an official county policy for which Hamilton County is liable under § 1983. This holding is wrong for at least two reasons. First, the Prosecutor’s response and the Deputies’ subsequent actions did not violate any constitutional right that existed at the time of the forcible entry. Second, no official county policy could have been created solely by an offhand telephone response from a busy County Prosecutor.

» — I

Petitioner s allegation of a constitutional violation rests exclusively on Steagald v. United States, 451 U. S. 204 (1981), decided four years after the entry here. In Steagald we held that an officer may not search for the subject of an arrest warrant in a third party’s home without first obtaining a search warrant, unless the search is consensual or justified by exigent circumstances. In 1977, the law in the Sixth Circuit was that a search warrant was not required in such situations if the police had an arrest warrant and reason to believe that the person to be arrested was within the home to be *493searched. United States v. McKinney, 379 F. 2d 259, 262-263 (1967). That view was shared by at least two other Circuits. See United States v. Gaultney, 606 F. 2d 540, 544-545 (CA5 1979); United States v. Harper, 550 F. 2d 610, 612-614 (CA10), cert. denied, 434 U. S. 837 (1977). Another Circuit had favored that view in dicta. See United States v. Manley, 632 F. 2d 978, 983 (CA2 1980). Thus, under the governing law in the applicable Circuit, uncontradicted by any opinion of this Court, the entry into petitioner’s office pursuant to an arrest warrant was not a violation of petitioner’s Fourth Amendment rights. .

The only way to transform this search — legitimate at the time — into a constitutional violation is to apply Steagald retroactively. This would not be a startling proposition if all that petitioner sought was retroactive application of a new rule of criminal law to a direct appeal from his criminal conviction.1 But petitioner seeks something very different— retroactive application of the new rule of criminal law announced in Steagald to his subsequent civil lawsuit. Even if one accepts the proposition that a new rule of criminal law should be applied retroactively to create a basis for civil liability under § 1983,2 existing principles of retroactivity for *494civil cases show that Steagald should not be applied retroactively to this action.

The leading case explaining the framework of analysis for civil retroactivity is Chevron Oil Co. v. Huson, 404 U. S. 97 (1971). Under Chevron, a court must assess: (i) whether the new decision “established] a new principle of law ... by overruling clear past precedent... or by deciding an issue of first impression whose resolution was not clearly foreshadowed,” id., at 106; (ii) “the prior history of the rule in question, its purpose and effect, and whether retrospective operation will further or retard its operation,” id., at 107; and (iii) the respective inequities of retroactive or nonretroactive application, ibid.

When viewed in light of these factors, retroactive application of Steagald is not justified. First, Steagald overruled past Courts of Appeals precedent, and the decision had not been foreshadowed in opinions of this Court. The governing law in three Federal Circuits permitted searches of third parties’ homes pursuant to an arrest warrant, see swpra, at 493, and earlier decisions of this Court arguably supported such searches.3 Second, the “purpose” of Steagald was to clarify the application of the Fourth Amendment to such searches, not to provide for money damages. Finally, retro*495active application of Steagald in this context would produce substantial inequitable results by imposing liability on local government units for law enforcement practices that were legitimate at the time they were undertaken. See Griffin v. Illinois, 351 U. S. 12, 26 (1956) (Frankfurter, J., concurring in judgment) (“We should not indulge in the fiction that the law now announced has always been the law . . .”)• Civil liability should not attach unless there was notice that a constitutional right was at risk. Procunier v. Navarette, 434 U. S. 555, 562 (1978).

We ought to be even more wary of applying new rules of Fourth Amendment law retroactively to civil cases than we are with new rules of civil law. The primary reason for imposing §1983 liability on local government units is deterrence, so that if there is any doubt about the constitutionality of their actions, officials will “err on the side of protecting citizens’ rights.” Owen v. City of Independence, 445 U. S. 622, 652 (1980). But law enforcement officials, particularly prosecutors, are in a much different position with respect to deterrence than other local government officials. Cf. Imbler v. Pachtman, 424 U. S. 409, 425 (1976). Their affirmative duty to enforce the law vigorously often requires them to take actions that legitimately intrude on individual liberties, often acting “under serious constraints of time and even information.” Ibid. While law enforcement officials, as much as any other official, ought to “err on the side of protecting citizens’ rights” when they have legitimate doubts about the constitutionality of their actions, they should not be deterred from doing their duty to enforce the criminal law when they have no such doubts. In this case, for example, Sixth Circuit law expressly authorized the Prosecutor’s decision. Because a court engages in the same balancing of interests in a Fourth Amendment case that is required, with much less deliberation, of law enforcement officials, they are justified in relying on the judgment of the applicable federal court. Under these circumstances, there was nothing that should *496have caused the officials to “harbor doubts about the lawfulness of their intended actions,” Owen, supra, at 652, and therefore nothing to deter.

Moreover, there is a significant cost to unwarranted deterrence of law enforcement officials. We recognized in Irnbler a strong state interest in “vigorous and fearless” prosecution, and found that to be “essential to the proper functioning of the criminal justice system.” 424 U. S., at 427-428. Those same general concerns apply to other law enforcement officials. Unwarranted deterrence has the undesirable effect of discouraging conduct that is essential to our justice system and protects the State’s interest in public safety. In that sense, this case is different from Owen. It is no answer to say that some officials are entitled either to absolute or qualified immunity. It ignores reality to say that if petitioner is successful in his $20-million suit it will not have a chilling effect on law enforcement practices in Hamilton County.4

For these reasons, Steagald should not be applied retroactively. Consequently, petitioner has no constitutional violation of which to complain. I therefore would affirm the decision of the Court of Appeals.5

*497II

Even if Steagald is applied retroactively, petitioner has failed to demonstrate the existence of an official policy for which Hamilton County can be liable. The action said to have created policy here was nothing more than a brief response to a single question over the telephone. The Deputy Sheriffs sought instructions concerning a situation that had never occurred before, at least in the memory of the participants. Ante, at 474-475. That in itself, and the fact that the Assistant Prosecutor had to obtain advice from the County Prosecutor, strongly indicate that no prior policy had been formed. Petitioner therefore argues that the County Prosecutor’s reaction in this case formed county policy. The sparse facts supporting petitioner’s theory — adopted by the Court today — do not satisfy the requirement in Monell v. New York City Dept. of Social Services, 436 U. S. 658, 691 (1978), that local government liability under §1983 be imposed only when the injury is caused by government policy.

A

Under Monell, local government units may be liable only when “the action that is alleged to be unconstitutional implements or executes a policy statement, ordinance, regulation, or decision officially adopted and promulgated by that body’s officers.” 436 U. S., at 690. This case presents the opportunity to define further what was meant in Monell by “official policy.” Proper resolution of the case calls for identification of the applicable principles for determining when policy is *498created. The Court today does not do this, but instead focuses almost exclusively on the status of the decisionmaker. Its reasoning is circular: it contends that policy is what policymakers make, and policymakers are those who have authority to make policy.

The Court variously notes that if a decision “is properly made by that government’s authorized decisionmakers, it surely represents an act of official government ‘policy’ as that term is commonly understood,” ante, at 481, and that “where action is directed by those who establish governmental policy, the municipality is equally responsible . . . ,” ibid. Thus, the Court’s test for determining the existence of policy focuses only on whether a decision was made “by the official or officials responsible for establishing final policy with respect to the subject matter in question.” Ante, at 483-484.

In my view, the question whether official policy — in any normal sense of the term — has been made in a particular case is not answered by explaining who has final authority to make policy. The question here is not “could the County Prosecutor make policy?” but rather, “did he make policy?” By focusing on the authority granted to the official under state law, the Court’s test fails to answer the key federal question presented. The Court instead turns the question into one of state law. Under a test that focuses on the authority of the decisionmaker, the Court has only to look to state law for the resolution of this case. Here the Court of Appeals found that “both the County Sheriff and the County Prosecutor [had authority under Ohio law to] establish county policy under appropriate circumstances.” Ante, at 484. Apparently that recitation of authority is all that is needed under the Court’s test because no discussion is offered to demonstrate that the Sheriff or the Prosecutor actually used that authority to establish official county policy in this case.

*499Moreover, the Court’s reasoning is inconsistent with Monell. Today’s decision finds that policy is established because a policymaking official made a decision on the telephone that was within the scope of his authority. The Court ignores the fact that no business organization or governmental unit makes binding policy decisions so cavalierly. The Court provides no mechanism for distinguishing those acts or decisions that cannot fairly be construed to create official policy from the normal process of establishing an official policy that would be followed by a responsible public entity. Thus, the Court has adopted in part what it rejected in Monell: local government units are now subject to respondeat superior liability, at least with respect to a certain category of employees, i. e., those with final authority to make policy. See Monell, 436 U. S., at 691; see also Oklahoma City v. Tuttle, 471 U. S. 808, 818 (1985) (rejecting theories akin to respondeat superior) (plurality opinion). The Court’s reliance on the status of the employee carries the concept of “policy” far beyond what was envisioned in Monell.

B

In my view, proper resolution of the question whether official policy has been formed should focus on two factors: (i) the nature of the decision reached or the action taken, and (ii) the process by which the decision was reached or the action was taken.

Focusing on the nature of the decision distinguishes between policies and mere ad hoc decisions. Such a focus also reflects the fact that most policies embody a rule of general applicability. That is the tenor of the Court’s statement in Monell that local government units are liable under § 1983 when the action that is alleged to be unconstitutional “implements or executes a policy statement, ordinance, regulation, or decision officially adopted and promulgated by that body’s officers.” 436 U. S., at 690. The clear implication is that policy is created when a rule is formed that applies to all simi*500lar situations — a “governing principle [or] plan.” Webster’s New Twentieth Century Dictionary 1392 (2d ed. 1979).6 When a rule of general applicability has been approved, the government has taken a position for which it can be held responsible.7

Another factor indicating that policy has been formed is the process by which the decision at issue was reached. Formal procedures that involve, for example, voting by elected officials, prepared reports, extended deliberation, or official records indicate that the resulting decisions taken “may fairly be said to represent official policy,” Monell, supra, at 694. Owen v. City of Independence, 445 U. S. 622 (1980), provides an example. The City Council met in a regularly scheduled meeting. One member of the Council made a motion to release to the press certain reports that cast an employee in a bad light. After deliberation, the Council passed the motion with no dissents and one abstention. Id., at 627-629. Although this official action did not establish a rule of general applicability, it is clear that policy was formed because of the process by which the decision was reached.

Applying these factors to the instant case demonstrates that no official policy was formulated. Certainly, no rule of general applicability was adopted. The Court correctly *501notes that the Sheriff “testified that the Department had no written policy respecting the serving of capiases on the property of third persons and that the proper response in any given situation would depend upon the circumstances.” Ante, at 474. Nor could he recall a specific instance in which entrance had been denied and forcibly gained. The Court’s result today rests on the implicit conclusion that the Prosecutor’s response — “go in and get them” — altered the prior case-by-case approach of the Department and formed a new rule to apply in all similar cases. Nothing about the Prosecutor’s response to the inquiry over the phone, nor the circumstances surrounding the response, indicates that such a rule of general applicability was formed.8

Similarly, nothing about the way the decision was reached indicates that official policy was formed. The prosecutor, without time for thoughtful consideration or consultation, simply gave an off-the-cuff answer to a single question. There was no process at all. The Court’s holding undercuts *502the basic rationale of Monell, and unfairly increases the risk of liability on the level of government least able to bear it. I dissent.

Justice O’Connor

announced the judgment of the Court and delivered an opinion, in which The Chief Justice, Justice White, and Justice Scalia join.

This case calls upon us to define the proper legal standard for determining when isolated decisions by municipal officials or employees may expose the municipality itself to liability under 42 U. S. C. § 1983.

I

The principal facts are not in dispute. Respondent James H. Praprotnik is an architect who began working for petitioner city of St. Louis in 1968. For several years, respondent consistently received favorable evaluations of his job performance, uncommonly quick promotions, and significant increases in salary. By 1980, he was serving in a management-level city planning position at petitioner’s Community Development Agency (CDA).

The Director of CDA, Donald Spaid, had instituted a requirement that the agency’s professional employees, including architects, obtain advance approval before taking on private clients. Respondent and other CDA employees ob*115jected to the requirement. In April 1980, respondent was suspended for 15 days by CDA’s Director of Urban Design, Charles Kindleberger, for having accepted outside employment without prior approval. Respondent appealed to the city’s Civil Service Commission, a body charged with reviewing employee grievances. Finding the penalty too harsh, the Commission reversed the suspension, awarded respondent backpay, and directed that he be reprimanded for having failed to secure a clear understanding of the rule.

The Commission’s decision was not well received by respondent’s supervisors at CDA. Kindleberger later testified that he believed respondent had lied to the Commission, and that Spaid was angry with respondent.

Respondent’s next two annual job performance evaluations were markedly less favorable than those in previous years. In discussing one of these evaluations with respondent, Kindleberger apparently mentioned his displeasure with respondent’s 1980 appeal to the Civil Service Commission. Respondent appealed both evaluations to the Department of Personnel. In each case, the Department ordered partial relief and was upheld by the city’s Director of Personnel or the Civil Service Commission.

In April 1981, a new Mayor came into office, and Donald Spaid was replaced as Director of CDA by Frank Hamsher. As a result of budget cuts, a number of layoffs and transfers significantly reduced the size of CDA and of the planning section in which respondent worked. Respondent, however, was retained.

In the spring of 1982, a second round of layoffs and transfers occurred at CDA. At that time, the city’s Heritage and Urban Design Commission (Heritage) was seeking approval to hire someone who was qualified in architecture and urban planning. Hamsher arranged with the Director of Heritage, Henry Jackson, for certain functions to be transferred from CDA to Heritage. This arrangement, which made it possible for Heritage to employ a relatively high-level “city planning *116manager,” was approved by Jackson’s supervisor, Thomas Nash. Hamsher then transferred respondent to Heritage to fill this position.

Respondent objected to the transfer, and appealed to the Civil Service Commission. The Commission declined to hear the appeal because respondent had not suffered a reduction in his pay or grade. Respondent then filed suit in Federal District Court, alleging that the transfer was unconstitutional. The city was named as a defendant, along with Kindleberger, Hamsher, Jackson (whom respondent deleted from the list before trial), and Deborah Patterson, who had succeeded Hamsher at CDA.

At Heritage, respondent became embroiled in a series of disputes with Jackson and Jackson’s successor, Robert Killen. Respondent was dissatisfied with the work he was assigned, which consisted of unchallenging clerical functions far below the level of responsibilities that he had previously enjoyed. At least one adverse personnel decision was taken against respondent, and he obtained partial relief after appealing that decision.

In December 1983, respondent was laid off from Heritage. The layoff was attributed to a lack of funds, and this apparently meant that respondent’s supervisors had concluded that they could create two lower level positions with the funds that were being used to pay respondent’s salary. Respondent then amended the complaint in his lawsuit to include a challenge to the layoff. He also appealed to the Civil Service Commission, but proceedings in that forum were postponed because of the pending lawsuit and have never been completed. Tr. Oral Arg. 31-32.

The case went to trial on two theories: (1) that respondent’s First Amendment rights had been violated through retaliatory actions taken in response to his appeal of his 1980 suspension; and (2) that respondent’s layoff from Heritage was carried out for pretextual reasons in violation of due process. The jury returned special verdicts exonerating *117each of the three individual defendants, but finding the city liable under both theories. Judgment was entered on the verdicts, and the city appealed.

A panel of the Court of Appeals for the Eighth Circuit found that the due process claim had been submitted to the jury on an erroneous legal theory and vacated that portion of the judgment. With one judge dissenting, however, the panel affirmed the verdict holding the city liable for violating respondent’s First Amendment rights. 798 F. 2d 1168 (1986). Only the second of these holdings is challenged here.

The Court of Appeals found that the jury had implicitly determined that respondent’s layoff from Heritage was brought about by an unconstitutional city policy. Id., at 1173. Applying a test under which a “policymaker” is one whose employment decisions are “final” in the sense that they are not subjected to de novo review by higher ranking officials, the Court of Appeals concluded that the city could be held liable for adverse personnel decisions taken by respondent’s supervisors. Id., at 1173-1175. In response to petitioner’s contention that the city’s personnel policies are actually set by the Civil Service Commission, the Court of Appeals concluded that the scope of review before that body was too “highly circumscribed” to allow it fairly to be said that the Commission, rather than the officials who initiated the actions leading to respondent’s injury, were the “final authority” responsible for setting city policy. Id., at 1175.

Turning to the question whether a rational jury could have concluded that respondent had been injured by an unconstitutional policy, the Court of Appeals found that respondent’s transfer from CDA to Heritage had been “orchestrated” by Hamsher, that the transfer had amounted to a “constructive discharge,” and that the injury had reached fruition when respondent was eventually laid off by Nash and Killen. Id., at 1175-1176, and n. 8. The court held that the jury’s verdict exonerating Hamsher and the other individual defendants could be reconciled with a finding of liability *118against the city because “the named defendants were not the supervisors directly causing the lay off, when the actual damages arose.” Id., at 1173, n. 3. Cf. Los Angeles v. Heller, 475 U. S. 796 (1986).

The dissenting judge relied on our decision in Pembaur v. Cincinnati, 475 U. S. 469 (1986). He found that the power to set employment policy for petitioner city of St. Louis lay with the Mayor and Aldermen, who were authorized to enact ordinances, and with the Civil Service Commission, whose function was to hear appeals from city employees who believed that their rights under the city’s Charter, or under applicable rules and ordinances, had not been properly respected. 798 F. 2d, at 1180. The dissent concluded that respondent had submitted no evidence proving that the Mayor and Aldermen, or the Commission, had established a policy of retaliating against employees for appealing from adverse personnel decisions. Id., at 1179-1181. The dissenting judge also concluded that, even if there were such a policy,, the record evidence would not support a finding that respondent was in fact transferred or laid off in retaliation for the 1980 appeal from his suspension. Id., at 1181-1182.

We granted certiorari, 479 U. S. 1029 (1987), and we now reverse.

II

We begin by addressing a threshold procedural issue. The second question presented in the petition for certiorari reads as follows:

“Whether the failure of a local government to establish an appellate procedure for the review of officials’ decisions which does not defer in substantial part to the original decisionmaker’s decision constitutes a delegation of authority to establish final government policy such that liability may be imposed on the local government on the basis of the decisionmaker’s act alone, when the act is neither taken pursuant to a rule of general applicability *119nor is a decision of specific application adopted as the result of a formal process?” Pet. for Cert. i.

Although this question was manifestly framed in light of the holding of the Court of Appeals, respondent argues that petitioner failed to preserve the question through a timely objection to the jury instructions under Federal Rule of Civil Procedure 51. Arguing that both parties treated the identification of municipal “policymakers” as a question of fact at trial, respondent emphasizes that the jury was given the following instruction, which was offered by the city itself:

“As a general principle, a municipality is not liable under 42 U. S. C. 1983 for the actions of its employees. However, a municipality may be held liable under 42 U. S. C. 1983 if the allegedly unconstitutional act was committed by an official high enough in the government so that his or her actions can be said to represent a government decision.” App. 113.

Relying on Oklahoma City v. Tuttle, 471 U. S. 808 (1985), and Springfield v. Kibbe, 480 U. S. 257 (1987), respondent contends that the jury instructions should be reviewed only for plain error, and that the jury’s verdict should be tested only for sufficiency of the evidence. Declining to defend the legal standard adopted by the Court of Appeals, respondent vigorously insists that the judgment should be affirmed on the basis of the jury’s verdict and petitioner’s alleged failure to comply with Rule 51.

Petitioner argues that it preserved the legal issues presented by its petition for certiorari in at least two ways. First, it filed a pretrial motion for summary judgment, or alternatively for judgment on the pleadings. In support of that motion, petitioner argued that respondent had failed to allege the existence of any impermissible municipal policy or of any facts that would indicate that such a policy existed. Second, petitioner filed a motion for directed verdict at the close of respondent’s case, renewed that motion at the close *120of all the evidence, and eventually filed a motion for judgment notwithstanding the verdict.

Respondent’s arguments do not bring our jurisdiction into question, and we must not lose sight of the fact, stressed in Tuttle, that the “decision to grant certiorari represents a commitment of scarce judicial resources with a view to deciding the merits of one or more of the questions presented in the petition.” 471 U. S., at 816. In Kibbe, it is true, the writ was dismissed in part because the petitioner sought to challenge a jury instruction to which it had not objected at trial. In the case before us, the focus of petitioner’s challenge is not on the jury instruction itself, but on the denial of its motions for summary judgment and a directed verdict. Although the same legal issue was raised both by those motions and by the jury instruction, “the failure to object to an instruction does not render the instruction the 'law of the case’ for purposes of appellate review of the denial of a directed verdict or judgment notwithstanding the verdict.” Kibbe, supra, at 264 (dissenting opinion) (citations omitted). Petitioner’s legal position in the District Court — that respondent had failed to establish an unconstitutional municipal policy — was consistent with the legal standard that it now advocates. It should not be surprising if petitioner’s arguments in the District Court were much less detailed than the arguments it now makes in response to the decision of the Court of Appeals. That, however, does not imply that petitioner failed to preserve the issue raised in its petition for certiorari. Cf. post, at 165-167 (Stevens, J., dissenting). Accordingly, we find no obstacle to reviewing the question presented in the petition for certiorari, a question that was very clearly considered, and decided, by the Court of Appeals.

We note, too, that petitioner has throughout this litigation been confronted with a legal landscape whose contours are “in a state of evolving definition and uncertainty.” Newport v. Fact Concerts, Inc., 453 U. S. 247, 256 (1981). We therefore do not believe that our review of the decision of the *121Court of Appeals, a decision raising a question that “is important and appears likely to recur in § 1983 litigation against municipalities,” id., at 257, will undermine the policy of judicial efficiency that underlies Rule 51. The definition of municipal liability manifestly needs clarification, at least in part to give lower courts and litigants a fairer chance to craft jury instructions that will not require scrutiny on appellate review.

Ill

A

Section 1 of the Ku Klux Act of 1871, Rev. Stat. § 1979, as amended, 42 U. S. C. § 1983, provides:

“Every person who, under color of any statute, ordinance, regulation, custom, or usage, of any State . . . , subjects, or causes to be subjected, any citizen of the United States or other person within the jurisdiction thereof to the deprivation of any rights, privileges, or immunities secured by the Constitution and laws, shall be liable to the party injured in an action at law, suit in equity, or other proper proceeding for redress. ...”

Ten years ago, this Court held that municipalities and other bodies of local government are “persons” within the meaning of this statute. Such a body may therefore be sued directly if it is alleged to have caused a constitutional tort through “a policy statement, ordinance, regulation, or decision officially adopted and promulgated by that body’s officers.” Monell v. New York City Dept. of Social Services, 436 U. S. 658, 690 (1978). The Court pointed out that § 1983 also authorizes suit “for constitutional deprivations visited pursuant to governmental ‘custom’ even though such a custom has not received formal approval through the body’s official decisionmaking channels.” Id., at 690-691. At the same time, the Court rejected the use of the doctrine of respondeat superior and concluded that municipalities could be held liable only when an injury was inflicted by a govern-*122merit’s “lawmakers or by those whose edicts or acts may fairly be said to represent official policy.” Id., at 694.

MonelVs rejection of respondeat superior, and its insistence that local governments could be held liable only for the results of unconstitutional governmental “policies,” arose from the language and history of § 1983. For our purposes here, the crucial terms of the statute are those that provide for liability when a government “subjects [a person], or causes [that person] to be subjected,” to a deprivation of constitutional rights. Aware that governmental bodies can act only through natural persons, the Court concluded that these governments should be held responsible when, and only when, their official policies cause their employees to violate another person’s constitutional rights. Reading the statute’s language in the light of its legislative history, the Court found that vicarious liability would be incompatible with the causation requirement set out on the face of § 1983. See id., at 691. That conclusion, like decisions that have widened the scope of § 1983 by recognizing constitutional rights that were unheard of in 1871, has been repeatedly reaffirmed. See, e. g., Owen v. City of Independence, 445 U. S. 622, 633, 655, n. 39 (1980); Polk County v. Dodson, 454 U. S. 312, 325 (1981); Tuttle, 471 U. S., at 818, and n. 5 (plurality opinion); id., at 828 (Brennan, J., concurring in part and concurring in judgment); Pembaur v. Cincinnati, 475 U. S., at 478-480, and nn. 7-8. Cf. Newport v. Fact Concerts, Inc., supra, at 259 (“[B]ecause the 1871 Act was designed to expose state and local officials to a new form of liability, it would defeat the promise of the statute to recognize any pre-existing immunity without determining both the policies that it serves and its compatibility with the purposes of § 1983”).

In Monell itself, it was undisputed that there had been an official policy requiring city employees to take actions that were unconstitutional under this Court’s decisions. Without attempting to draw the line between actions taken pursuant to official policy and. the independent actions of employees *123and agents, the Monell Court left the “full contours” of municipal liability under § 1983 to be developed further on “another day.” 436 U. S., at 695.

In the years since Monell was decided, the Court has considered several cases involving isolated acts by government officials and employees. We have assumed that an unconstitutional governmental policy could be inferred from a single decision taken by the highest officials responsible for setting policy in that area of the government’s business. See, e. g., Owen v. City of Independence, supra; Newport v. Fact Concerts, Inc., 453 U. S. 247 (1981). Cf. Pembaur, supra, at 480. At the other end of the spectrum, we have held that an unjustified shooting by a police officer cannot, without more, be thought to result from official policy. Tuttle, 471 U. S., at 821 (plurality opinion); id., at 830-831, and n. 5 (Brennan, J., concurring in part and concurring in judgment). Cf. Kibbe, 480 U. S., at 260 (dissenting opinion).

Two Terms ago, in Pembaur, supra, we undertook to define more precisely when a decision on a single occasion may be enough to establish an unconstitutional municipal policy. Although the Court was unable to settle on a general formulation, Justice Brennan’s opinion articulated several guiding principles. First, a majority of the Court agreed that municipalities may be held liable under § 1983 only for acts for which the municipality itself is actually responsible, “that is, acts which the municipality has officially sanctioned or ordered.” Id., at 480. Second, only those municipal officials who have “final policymaking authority” may by their actions subject the government to § 1983 liability. Id., at 483 (plurality opinion). Third, whether a particular official has “final policymaking authority” is a question of state law. Ibid, (plurality opinion). Fourth, the challenged action must have been taken pursuant to a policy adopted by the official or officials responsible under state law for making policy in that area of the city’s business. Id., at 482-483, and n. 12 (plurality opinion).

*124The Courts of Appeals have already diverged in their interpretations of these principles. Compare, for example, Williams v. Butler, 802 F. 2d 296, 299-302 (CA8 1986) (en banc), cert. pending sub nom. Little Rock v. Williams, No. 86-1049, with Jett v. Dallas Independent School Dist., 798 F. 2d 748, 759-760 (CA5 1986) (dictum). Today, we set out again to clarify the issue that we last addressed in Pembaur.

B

We begin by reiterating that the identification of policy-making officials is a question of state law. “Authority to make municipal policy may be granted directly by a legislative enactment or may be delegated by an official who possesses such authority, and of course, whether an official had final policymaking authority is a question of state law.” Pembaur v. Cincinnati, supra, at 483 (plurality opinion).1 Thus the identification of policymaking officials is not a question of federal law, and it is not a question of fact in the usual sense. The States have extremely wide latitude in determining the form that local government takes, and local preferences have led to a profusion of distinct forms. Among the many kinds of municipal corporations, political subdivisions, and special districts of all sorts, one may expect to find a rich variety of ways in which the power of govern*125ment is distributed among a host of different officials and official bodies. See generally C. Rhyne, The Law of Local Government Operations §§1.3-1.7 (1980). Without attempting to canvass the numberless factual scenarios that may come to light in litigation, we can be confident that state law (which may include valid local ordinances and regulations) will always direct a court to some official or body that has the responsibility for making law or setting policy in any given area of a local government’s business.2

We are not, of course, predicting that state law will always speak with perfect clarity. We have no reason to suppose, *126however, that federal courts will face greater difficulties here than those that they routinely address in other contexts. We are also aware that there will be cases in which policy-making responsibility is shared among more than one official or body. In the case before us, for example, it appears that the Mayor and Aldermen are authorized to adopt such ordinances relating to personnel administration as are compatible with the City Charter. See St. Louis City Charter, Art. XVIII, § 7(b), App. 62-63. The Civil Service Commission, for its part, is required to “prescribe . . . rules for the administration and enforcement of the provisions of this article, and of any ordinance adopted in pursuance thereof, and not inconsistent therewith.” §7(a), App. 62. Assuming that applicable law does not make the decisions of the Commission reviewable by the Mayor and Aldermen, or vice versa, one would have to conclude that policy decisions made either by the Mayor and Aldermen or by the Commission would be attributable to the city itself. In any event, however, a federal court would not be justified in assuming that municipal policymaking authority lies somewhere other than where the applicable law purports to put it. And certainly there can be no justification for giving a jury the discretion to determine which officials are high enough in the government that their actions can be said to represent a decision of the government itself.

As the plurality in Pembaur recognized, special difficulties can arise when it is contended that a municipal policymaker has delegated his policymaking authority to another official. 475 U. S., at 482-483, and n. 12. If the mere exercise of discretion by an employee could give rise to a constitutional violation, the result would be indistinguishable from respondeat superior liability. If, however, a city’s lawful policymakers could insulate the government from liability simply by delegating their policymaking authority to others, § 1983 could not serve its intended purpose. It may not be possible to draw an *127elegant line that will resolve this conundrum, but certain principles should provide useful guidance.

First, whatever analysis is used to identify municipal policymakers, egregious attempts by local governments to insulate themselves from liability for unconstitutional policies are precluded by a separate doctrine. Relying on the language of § 1983, the Court has long recognized that a plaintiff may be able to prove the existence of a widespread practice that, although not authorized by written law or express municipal policy, is “so permanent and well settled as to constitute a ‘custom or usage’ with the force of law.” Adickes v. S. H. Kress & Co., 398 U. S. 144, 167-168 (1970). That principle, which has not been affected by Monell or subsequent cases, ensures that most deliberate municipal evasions of the Constitution will be sharply limited.

Second, as the Pembaur plurality recognized, the authority to make municipal policy is necessarily the authority to make final policy. 475 U. S., at 481-484. When an official’s discretionary decisions are constrained by policies not of that official’s making, those policies, rather than the subordinate’s departures from them, are the act of the municipality. Similarly, when a subordinate’s decision is subject to review by the municipality’s authorized policymakers, they have retained the authority to measure the official’s conduct for conformance with their policies. If the authorized policymakers approve a subordinate’s decision and the basis for it, their ratification would be chargeable to the municipality because their decision is final.

C

Whatever refinements of these principles may be suggested in the future, we have little difficulty concluding that the Court of Appeals applied an incorrect legal standard in this case. In reaching this conclusion, we do not decide whether the First Amendment forbade the city to retaliate against respondent for having taken advantage of the grievance mechanism in 1980. Nor do we decide whether there *128was evidence in this record from which a rational jury could conclude either that such retaliation actually occurred or that respondent suffered any compensable injury from whatever retaliatory action may have been taken. Finally, we do not address petitioner’s contention that the jury verdict exonerating the individual defendants cannot be reconciled with the verdict against the city. Even assuming that all these issues were properly resolved in respondent’s favor, we would not be able to affirm the decision of the Court of Appeals.

The city cannot be held liable under § 1983 unless respondent proved the existence of an unconstitutional municipal policy. Respondent does not contend that anyone in city government ever promulgated, or even articulated, such a policy. Nor did he attempt to prove that such retaliation was ever directed against anyone other than himself. Respondent contends that the record can be read to establish that his supervisors were angered by his 1980 appeal to the Civil Service Commission; that new supervisors in a new administration chose, for reasons passed on through some informal means, to retaliate against respondent two years later by transferring him to another agency; and that this transfer was part of a scheme that led, another year and a half later, to his layoff. Even if one assumes that all this was true, it says nothing about the actions of those whom the law established as the makers of municipal policy in matters of personnel administration. The Mayor and Aldermen enacted no ordinance designed to retaliate against respondent or against similarly situated employees. On the contrary, the city established an independent Civil Service Commission and empowered it to review and correct improper personnel actions. Respondent does not deny that his repeated appeals from adverse personnel decisions repeatedly brought him at least partial relief, and the Civil Service Commission never so much as hinted that retaliatory transfers or layoffs were permissible. Respondent points to no evidence indicating that the Commission delegated to anyone its final authority to *129interpret and enforce the following policy set out in Article XVIII of the city’s Charter, §2(a), App. 49:

“Merit and fitness. All appointments and promotions to positions in the service of the city and all measures for the control and regulation of employment in such positions, and separation therefrom, shall be on the sole basis of merit and fitness . . . .”

The Court of Appeals concluded that “appointing authorities,” like Hamsher and Killen, who had the authority to initiate transfers and layoffs, were municipal “policymakers.” The court based this conclusion on its findings (1) that the decisions of these employees were not individually reviewed for “substantive propriety” by higher supervisory officials; and (2) that the Civil Service Commission decided appeals from such decisions, if at all, in a circumscribed manner that gave substantial deference to the original decisionmaker. 798 F. 2d, at 1174-1175. We find these propositions insufficient to support the conclusion that Hamsher and Killen were authorized to establish employment policy for the city with respect to transfers and layoffs. To the contrary, the City Charter expressly states that the Civil Service Commission has the power and the duty:

“To consider and determine any matter involved in the administration and enforcement of this [Civil Service] article and the rules and ordinances adopted in accordance therewith that may be referred to it for decision by the director [of personnel], or on appeal by any appointing authority, employe, or taxpayer of the city, from any act of the director or of any appointing authority. The decision of the commission in all such matters shall be final, subject, however, to any right of action under any law of the state or of the United States.” St. Louis City Charter, Art. XVIII, § 7(d), App. 63.

This case therefore resembles the hypothetical example in Pembaur: “[I]f [city] employment policy was set by the *130[Mayor and Aldermen and by the Civil Service Commission], only [those] bodies’] decisions would provide a basis for [city] liability. This would be true even if the [Mayor and Aider-men and the Commission] left the [appointing authorities] discretion to hire and fire employees and [they] exercised that discretion in an unconstitutional manner . . . .” 475 U. S., at 483, n. 12. A majority of the Court of Appeals panel determined that the Civil Service Commission’s review of individual employment actions gave too much deference to the decisions of appointing authorities like Hamsher and Killen. Simply going along with discretionary decisions made by one’s subordinates, however, is not a delegation to them of the authority to make policy. It is equally consistent with a presumption that the subordinates are faithfully attempting to comply with the policies that are supposed to guide them. It would be a different matter if a particular decision by a subordinate was cast in the form of a policy statement and expressly approved by the supervising policymaker. It would also be a different matter if a series of decisions by a subordinate official manifested a “custom or usage” of which the supervisor must have been aware. See supra, at 127. In both those cases, the supervisor could realistically be deemed to have adopted a policy that happened to have been formulated or initiated by a lower ranking official. But the mere failure to investigate the basis of a subordinate’s discretionary decisions does not amount to a delegation of policymaking authority, especially where (as here) the wrongfulness of the subordinate’s decision arises from a retaliatory motive or other unstated rationale. In such circumstances, the purposes of § 1983 would not be served by treating a subordinate employee’s decision as if it were a reflection of municipal policy.

Justice Brennan’s opinion, concurring in the judgment, finds implications in our discussion that we do not think necessary or correct. See post, at 142-147. We nowhere say or imply, for example, that “a municipal charter’s precatory *131admonition against discrimination or any other employment practice not based on merit and fitness effectively insulates the municipality from any liability based on acts inconsistent with that policy.” Post, at 145, n. 7. Rather, we would respect the decisions, embodied in state and local law, that allocate policymaking authority among particular individuals and bodies. Refusals to carry out stated policies could obviously help to show that a municipality’s actual policies were different from the ones that had been announced. If such a showing were made, we would be confronted with a different case than the one we decide today.

Nor do we believe that we have left a “gaping hole” in § 1983 that needs to be filled with the vague concept of “de facto final policymaking authority.” Post, at 144. Except perhaps as a step towards overruling Monell and adopting the doctrine of respondeat superior, ad hoc searches for officials possessing such “defacto” authority would serve primarily to foster needless unpredictability in the application of § 1983.

IV

We cannot accept either the Court of Appeals’ broad definition of municipal policymakers or respondent’s suggestion that a jury should be entitled to define for itself which officials’ decisions should expose a municipality to liability. Respondent has suggested that the record will support an inference that policymaking authority was in fact delegated to individuals who took retaliatory action against him and who were not exonerated by the jury. Respondent’s arguments appear to depend on a legal standard similar to the one suggested in Justice Stevens’ dissenting opinion, post, at 171, which we do not accept. Our examination of the record and state law, however, suggests that further review of this case may be warranted in light of the principles we have discussed. That task is best left to the Court of Appeals, which will be free to invite additional briefing and argument if necessary. Accordingly, the decision of the Court of Appeals is *132reversed, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

Justice Brennan,

with whom

Despite its somewhat confusing procedural background, this case at bottom presents a relatively straightforward question: whether respondent’s supervisor at the Community Development Agency, Frank Hamsher, possessed the authority to establish final employment policy for the city of St. Louis such that the city can be held liable under 42 U. S. C. § 1983 for Hamsher’s allegedly unlawful decision to transfer respondent to a dead-end job. Applying the test set out two Terms ago by the plurality in Pembaur v. Cincinnati, 475 U. S. 469 (1986), I conclude that Hamsher did not possess such authority and I therefore concur in the Court’s judgment reversing the decision below. I write separately, however, because I believe that the commendable desire of today’s plurality to “define more precisely when a decision on a single occasion may be enough” to subject a municipality to § 1983 liability, ante, at 123, has led it to embrace a theory of municipal liability that is both unduly narrow and unrealistic, and one that ultimately would permit municipalities to insulate themselves from liability for the acts of all but a small minority of actual city policymakers.

I

Respondent James H. Praprotnik worked for petitioner city of St. Louis for 15 years. A licensed architect, he began his career in 1968 as city planner and by 1980 had risen to a mid-level management position in the city’s Community Development Agency (CDA), garnering consistently high job evaluations, substantial pay raises, and rapid promotions *133during the intervening 12 years. 1980, however, marked the turning point in respondent’s fortunes as a civil servant. In April of that year, his supervisor, Charles Kindleberger, suspended him for 15 days for failing to comply with a secondary employment policy that required all city professionals to obtain prior approval before undertaking any outside work. Respondent, who had objected to the policy since the head of the agency, CDA Director Donald Spaid, first announced it in 1978, appealed the suspension to the city’s Civil Service Commission (CSC), arguing that the advance approval requirement was an improper invasion of his privacy and that in any event he had consistently complied with it. Although the CSC apparently did not question the validity of the policy, it found the penalty excessive, and therefore directed respondent’s supervisors to reinstate him with backpay and to issue a letter of reprimand in lieu of the suspension.

Testimony at the trial below revealed that neither Spaid nor Kindleberger was pleased with respondent’s actions, and that Spaid in particular was “very down on” respondent for his testimony before the CSC. 3 Record 1-54 to 1-55, 5 id., at 3-237. In October 1980, just before the CSC rendered its decision, Kindleberger gave respondent an overall rating of “good” for the year, but recommended a two-step decrease in his salary. Kindleberger, who had just six months earlier proposed raising respondent’s salary two grades, justified the reduction as part of a citywide pay scale reorganization. Respondent, however, viewed the recommendation as retaliation for his CSC appeal and petitioned the Department of Personnel for relief; the Department, which considers initial challenges to all performance ratings, granted partial relief, approving a one-step reduction, and the CSC affirmed this disposition on final appeal.

The following year witnessed a change in city administrations and the arrival of Frank Hamsher, who succeeded Spaid as CDA Director. Kindleberger, however, remained the supervisor responsible for respondent’s performance *134evaluation, and in October 1981 he rated respondent merely “adequate” overall. A confidential memorandum from one of respondent’s superiors to Kindleberger explained that respondent did not get along well with others, citing as an example respondent’s prior difficulties with former Director Spaid. Respondent, who had.previously never received a rating of less than “good,” again appealed to the Department of Personnel, which again ordered partial relief.

Six months later CDA underwent major budget and staff reductions and, as part of the resulting reorganization, Director Hamsher proposed transferring respondent’s duties to the Heritage and Urban Design Commission (Heritage) and consolidating his functions with those of a vacant position at Heritage. Although there was testimony indicating that Heritage Commissioner Henry Jackson thought the transfer unnecessary, both Jackson and his superior, Director of Public Safety Thomas Nash, agreed to the consolidation, and the Director of Personnel formally approved the proposal. Respondent objected to the move and appealed to the CSC, but the CSC declined to review the decision, reasoning that because Heritage classified the consolidated position at the same grade as respondent’s former job, the transfer was merely “lateral” and respondent had therefore suffered no “adverse” employment action. Thereafter, respondent filed this § 1983 suit against the city, Kindleberger, Hamsher, and Hamsher’s successor at CDA, Deborah Patterson, alleging that the transfer violated his constitutional rights.1

In the meantime, Jackson took over many of the architectural tasks CDA had ostensibly transferred to the new position and assigned respondent mainly clerical duties, an arrangement the latter found highly unsatisfactory. In November 1982, Jackson rated respondent “inadequate” overall and recommended a one-step reduction in his salary, as well *135as an overall reduction in the classification of his position. Respondent successfully appealed his performance rating to the Personnel Department, which again granted partial relief. Nonetheless, in March 1983 his position was substantially downgraded and by the summer of that year Jackson’s successor at Heritage, Robert Killen, proposed abolishing the position altogether. In December 1983, Killen carried through on his plan and, with the approval of Public Safety Director Nash, laid respondent off. Respondent amended his complaint in the District Court to reflect the layoff and simultaneously appealed the action to the CSC, but the CSC stayed its proceedings in light of the pendency of this lawsuit.

At trial, respondent sought to prove that the individual defendants had transferred him and eventually laid him off in retaliation for his use of the city’s grievance machinery, thereby violating his First Amendment and due process rights. For its part, the city contended that the individual defendants were not personally responsible for the alleged ills that had befallen respondent. Conspicuous by their absence, city counsel argued, were Donald Spaid, whose displeasure over respondent’s testimony before the CSC was allegedly the motivating force behind respondent’s first proposed grade reduction and allegedly infected later performance evaluations; Robert Killen, who initiated and ultimately authorized the elimination of respondent’s position at Heritage; and Thomas Nash, who approved the layoff. Respondent’s counsel, however, defended the choice of defendants as those “primarily responsible” for the constitutional deprivations. 6 id., at 4-56.

The District Court instructed the jury that generally a city is not liable under § 1983 for the acts of its employees, but that it may be held to answer for constitutional wrongs “committed by an official high enough in the government so that his or her actions can be said to represent a government decision.” App. 113. In a lengthy and involved instruction, the court further advised the jury that it must find in favor of *136respondent, and against the individual defendants, if it found six facts to be true, one of which was that “Hamsher and Kindleberger were personally involved in causing [respondent’s] transfer and/or layoff.” Id., at 118. The jury exonerated the three individual defendants, but awarded respondent $15,000 on each of his constitutional claims against petitioner.

The Court of Appeals for the Eighth Circuit vacated the judgment entered on respondent’s due process claim (a ruling not at issue here) but affirmed the judgment as to the First Amendment claim. 798 F. 2d 1168 (1986). With respect to this latter claim, the court reasoned that the city could be held accountable for an improperly motivated transfer and layoff if it had delegated to the responsible officials, either directly or indirectly, the authority to act on behalf of the city, and if the decisions made within the scope of this delegated authority were essentially final. Applying this test, the court noted that under the City Charter, “appointing authorities,” or department heads, such as Hamsher, could undertake transfers and layoffs subject only to the approval of the Director of Personnel, who undertook no substantive review of such decisions and simply conditioned his approval on formal compliance with city procedures. ' Moreover, because the CSC engaged in highly circumscribed and deferential review of layoffs and, at least so far as this case reveals, no review whatever of lateral transfers, the court concluded that an appointing authority’s transfer and layoff decisions were final. Id., at 1174-1175.

Having found that Hamsher was a final policymaker whose acts could subject petitioner to § 1983 liability, the court determined that the jury had ample evidence from which it could find that Hamsher transferred respondent in retaliation for the latter’s exercise of his First Amendment rights, and that the transfer in turn precipitated respondent’s layoff. This constructive discharge theory, the majority found, also reconciled the jury’s apparently inconsistent verdicts: the *137jury could have viewed Hamsher’s unlawful motivation as the proximate cause of respondent’s dismissal but, because Nash and Killen administered the final blows, it could have concluded that Hamsher, Kindleberger, and Patterson were not “personally involved” in the layoff as required by the instructions; accordingly, the jury could have reasonably exonerated the individual defendants while finding the city liable. Id., at 1176, and n. 8.2

II

In light of the jury instructions below, the central question before us is whether the city delegated to CDA Director Frank Hamsher the authority to establish final employment policy for the city respecting transfers. For if it did not, then his allegedly unlawful decision to move respondent to an unfulfilling, dead-end position is simply not an act for which the city can be held responsible under § 1983. I am constrained to conclude that Hamsher possessed no such policy-making power here, and that, on the contrary, his allegedly retaliatory act simply constituted an abuse of the discretionary authority the city had entrusted to him.

The scope of Hamsher’s authority with respect to transfers derives its significance from our determination in Monell v. New York City Dept. of Social Services, 436 U. S. 658 (1978), that a municipality is not liable under §1983 for each and every wrong committed by its employees. In rejecting the concept of vicarious municipal liability, we emphasized that *138“the touchstone of the § 1983 action against a government body is an allegation that official policy is responsible for the deprivation of rights protected by the Constitution.” Id., at 690. More recently we have explained that the touchstone of “official policy” is designed “to distinguish acts of the municipality from acts of employees of the municipality, and thereby make clear that municipal liability is limited to action for which the municipality is actually responsible.” Pembaur v. Cincinnati, 475 U. S., at 479-480 (emphasis in original).

Municipalities, of course, conduct much of the business of governing through human agents. Where those agents act in accordance with formal policies, or pursuant to informal practices “so permanent and well settled as to constitute a ‘custom or usage’ with the force of law,” Adickes v. S. H. Kress & Co., 398 U. S. 144, 167-168 (1970), we naturally ascribe their acts to the municipalities themselves and hold the latter responsible for any resulting constitutional deprivations. Monell, which involved a challenge to a citywide policy requiring all pregnant employees to take unpaid leave after their fifth- month of pregnancy, was just such a case. Nor have we ever doubted that a single decision of a city’s properly constituted legislative body is a municipal act capable of subjecting the city to liability. See, e. g., Newport v. Fact Concerts, Inc., 453 U. S. 247 (1981) (City Council canceled concert permit for content-based reasons); Owen v. City of Independence, 445 U. S. 622 (1980) (City Council passed resolution firing Police Chief without any pretermination hearing). In these cases we neither required nor, as the plurality suggests, assumed that these decisions reflected generally applicable “policies” as that term is commonly understood, because it was perfectly obvious that the actions of the municipalities’ policymaking organs, whether isolated or not, were properly charged to the municipalities themselves.*1393 And, in Pembaur we recognized that “the power to establish policy is no more the exclusive province of the legislature at the local level than at the state or national level,” 475 U. S., at 480, and that the isolated decision of an executive municipal policymaker, therefore, could likewise give rise to municipal liability under § 1983.

In concluding that Frank Hamsher was a policymaker, the Court of Appeals relied on the fact that the city had delegated to him “the authority, either directly or indirectly, to act on [its] behalf,” and that his decisions within the scope of this delegated authority were effectively final. 798 F. 2d, at 1174. In Pembaur, however, we made clear that a municipality is not liable merely because the official who inflicted the constitutional injury had the final authority to act on its behalf; rather, as four of us explained, the official in question must possess “final authority to establish municipal policy with respect to the [challenged] action.” 475 U. S., at 481. Thus, we noted, “[t]he fact that a particular official — even a policymaking official — has discretion in the exercise of particular functions does not, without more, give rise to munici*140pal liability based on an exercise of that discretion.” Id., at 481-482. By way of illustration, we explained that if, in a given county, the Board of County Commissioners established county employment policy and delegated to the County Sheriff alone the discretion to hire and fire employees, the county itself would not be liable if the Sheriff exercised this authority in an unconstitutional manner, because “the decision to act unlawfully would not be a decision of the Board.” Id., at 483, n. 12. We pointed out, however, that in that same county the Sheriff could be the final policymaker in other areas, such as law enforcement practices, and that if so, his or her decisions in such matters could give rise to municipal liability. Ibid. In short, just as in Owen and Fact Concerts we deemed it fair to hold municipalities liable for the isolated, unconstitutional acts of their legislative bodies, regardless of whether those acts were meant to establish generally applicable “policies,” so too in Pembaur four of us concluded that it is equally appropriate to hold municipalities accountable for the isolated constitutional injury inflicted by an executive final municipal policymaker, even though the decision giving rise to the injury is not intended to govern future situations. In either case, as long as the contested decision is made in an area over which the official or legislative body could establish a final policy capable of governing future municipal conduct, it is both fair and consistent with the purposes of § 1983 to treat the decision as that of the municipality itself, and to hold it liable for the resulting constitutional deprivation.

In my view, Pembaur controls this case. As an “appointing authority,” Hamsher was empowered under the City Charter to initiate lateral transfers such as the one challenged here, subject to the approval of both the Director of Personnel and the appointing authority of the transferee agency. The Charter, however, nowhere confers upon *141agency heads any authority to establish city policy, final or otherwise, with respect to such transfers. Thus, for example, Hamsher was not authorized to promulgate binding guidelines or criteria governing how or when lateral transfers were to be accomplished. Nor does the record reveal that he in fact sought to exercise any such authority in these matters. There is no indication, for example, that Hamsher ever purported to institute or announce a practice of general applicability concerning transfers. Instead, the evidence discloses but one transfer decision — the one involving respondent — which Hamsher ostensibly undertook pursuant to a citywide program of fiscal restraint and budgetary reductions. At most, then, the record demonstrates that Hamsher had the authority to determine how best to effectuate a policy announced by his superiors, rather than the power to establish that policy. Like the - hypothetical Sheriff in Pembaur’s n. 12, Hamsher had discretionary authority to transfer CDA employees laterally; that he may have used this authority to punish respondent for the exercise of his First Amendment rights does not, without more, render the city liable for respondent’s resulting constitutional injury.4 The court below did not suggest that either Killen or Nash, who together orchestrated respondent’s ultimate layoff, *142shared Hamsher’s constitutionally impermissible animus. Because the court identified only one unlawfully motivated municipal employee involved in respondent’s transfer and layoff, and because that employee did not possess final policy-making authority with respect to the contested decision,5 the city may not be held accountable for any constitutional wrong respondent may have suffered.

Ill

These determinations, it seems to me, are sufficient to dispose of this case, and I therefore think it unnecessary to decide, as the plurality does, who the actual policymakers in St. Louis are. I question more than the mere necessity of these determinations, however, for I believe that in the course of passing on issues not before us, the plurality announces legal principles that are inconsistent with our earlier cases and unduly restrict the reach of § 1983 in cases involving municipalities.

The plurality begins its assessment of St. Louis’ power structure by asserting that the identification of policymaking officials is a question of state law, by which it means that the question is neither one of federal law nor of fact, at least “not ... in the usual sense.” See ante, at 124. Instead, the plurality explains., courts are to identify municipal policymakers *143by referring exclusively to applicable state statutory law. Ibid. Not surprisingly, the plurality cites no authority for this startling proposition, nor could it, for we have never suggested that municipal liability should be determined in so formulaic and unrealistic a fashion. In any case in which the policymaking authority of a municipal tortfeasor is in doubt, state law will naturally be the appropriate starting point, but ultimately the factfinder must determine where such policymaking authority actually resides, and not simply “where the applicable law purports to put it.” Ante, at 126. As the plurality itself acknowledges, local governing bodies may take myriad forms. We in no way slight the dignity of municipalities by recognizing that in not a few of them real and apparent authority may diverge, and that in still others state statutory law will simply fail to disclose where such authority ultimately rests. Indeed, in upholding the Court of Appeals’ determination in Pembaur that the County Prosecutor was a policymaking official with respect to county law enforcement practices, a majority of this Court relied on testimony which revealed that the County Sheriff’s office routinely forwarded certain matters to the Prosecutor and followed his instructions in those areas. See 475 U. S., at 485; ibid. (White, J., concurring); id., at 491 (O’Connor, J., concurring). While the majority splintered into three separate camps on the ultimate theory of municipal liability, and the case generated five opinions in all, not a single Member of the Court suggested that reliance on such extrastatutory evidence of the county’s actual allocation of policymaking authority was in any way improper. Thus, although I agree with the plurality that juries should not be given open-ended “discretion to determine which officials are high enough in the government that their actions can be said to represent a decision of the government itself,” ante, at 126 (emphasis added), juries can and must find the predicate facts necessary to a determination whether a given official possesses final policymaking authority. While the jury instructions in this case were regrettably vague, the plurality’s solution tosses *144the baby out with the bath water. The identification of municipal policymakers is an essentially factual determination “in the usual sense,” and is therefore rightly entrusted to a properly instructed jury.

Nor does the “custom or usage” doctrine adequately compensate for the inherent inflexibility of a rule that leaves the identification of policymakers exclusively to state statutory law. That doctrine, under which municipalities and States can be held liable for unconstitutional practices so well settled and permanent that they have the force of law, see Adickes v. S. H. Kress & Co., 398 U. S., at 167, has little if any bearing on the question whether a city has delegated de facto final policymaking authority to a given official. A city practice of delegating final policymaking authority to a subordinate or mid-level official would not be unconstitutional in and of itself, and an isolated unconstitutional act by an official entrusted with such authority would obviously not amount to a municipal “custom or usage.” Under Pembaur, of course, such an isolated act should give rise to municipal liability. Yet a case such as this would fall through the gaping hole the plurality’s construction leaves in § 1983, because state statutory law would not identify the municipal actor as a policy-making official, and a single constitutional deprivation, by definition, is not a well-settled and permanent municipal practice carrying the force of law.6

For these same reasons, I cannot subscribe to the plurality’s narrow and overly rigid view of when a municipal official’s policymaking authority is “final.” Attempting to place a gloss on Pembaur’s finality requirement, the plurality suggests that whenever the decisions of an official are subject to *145some form of review — however limited — that official’s decisions are nonfinal. Under the plurality’s theory, therefore, even where an official wields policymaking authority with respect to a challenged decision, the city would not be liable for that official’s policy decision unless reviewing officials affirmatively approved both the “decision and the basis for it.” Ante, at 127. Reviewing officials, however, may as a matter of practice never invoke their plenary oversight authority, or their review powers may be highly circumscribed. See n. 4, supra. Under such circumstances, the subordinate’s decision is in effect the final municipal pronouncement on the subject. Certainly a § 1983 plaintiff is entitled to place such considerations before the jury, for the law is concerned not with the niceties of legislative draftsmanship but with the realities of municipal decisionmaking, and any assessment of a municipality’s actual power structure is necessarily a factual and practical one.7

Accordingly, I cannot endorse the plurality’s determination, based on nothing more than its own review of the City Charter, that the Mayor, the Aldermen, and the CSC are the only policymakers for the city of St. Louis. While these offi*146cials may well have policymaking authority, that hardly ends the matter; the question before us is whether the officials responsible for respondent’s allegedly unlawful transfer were final policymakers. As I have previously indicated, I do not believe that CDA Director Frank Hamsher possessed any policymaking authority with respect to lateral transfers and thus I do not believe that his allegedly improper decision to transfer respondent could, without more, give rise to municipal liability. Although the plurality reaches the same result, it does so by reasoning that because others could have reviewed the decisions of Hamsher and Killen, the latter officials simply could not have been final policymakers.

This analysis, however, turns a blind eye to reality, for it ignores not only the lower court’s determination, nowhere disputed, that CSC review was highly circumscribed and deferential, but also the fact that in this very case the CSC refused to judge the propriety of Hamsher’s transfer decision because a lateral transfer was not an “adverse” employment action falling within its jurisdiction. Nor does the plurality account for the fact that Hamsher’s predecessor, Donald Spaid, promulgated what the city readily acknowledges was a binding policy regarding secondary employment;8 although the CSC ultimately modified the sanctions respondent suffered as a result of his apparent failure to comply with that policy, the record is devoid of any suggestion that the CSC reviewed the substance or validity of the policy itself. Under the plurality’s analysis, therefore, even the hollowest promise of review is sufficient to divest all city officials save the mayor and governing legislative body of final policy-making authority. While clarity and ease of application may *147commend such a rule, we have remained steadfast in our conviction that Congress intended to hold municipalities accountable for those constitutional injuries inflicted not only by their lawmakers, but also “by those whose edicts or acts may fairly be said to represent official policy.” Monell, 436 U. S., at 694. Because the plurality’s mechanical “finality” test is fundamentally at odds with the pragmatic and factual inquiry contemplated by Monell, I cannot join what I perceive to be its unwarranted abandonment of the traditional factfinding process in § 1983 actions involving municipalities.

Finally, I think it necessary to emphasize that despite certain language in the plurality opinion suggesting otherwise, the Court today need not and therefore does not decide that a city can only be held liable under § 1983 where the plaintiff “prove[s] the existence of an unconstitutional municipal policy.” See ante, at 128. Just last Term, we left open for the second time the question whether a city can be subjected to liability for a policy that, while not unconstitutional in and of itself, may give rise to constitutional deprivations. See Springfield v. Kibbe, 480 U. S. 257 (1987); see also Oklahoma City v. Tuttle, 471 U. S. 808 (1985). That question is certainly not presented by this case, and nothing we say today forecloses its future consideration.

IV

For the reasons stated above, I concur in the judgment of the Court reversing the decision below and remanding the case so that the Court of Appeals may determine whether respondent’s layoff resulted from the actions of any improperly motivated final policymakers.

Justice Stevens,

dissenting.

If this case involved nothing more than a personal vendetta between a municipal employee and his superiors, it would be quite wrong to impose liability on the city of St. Louis. In fact, however, the jury found that top officials in the city administration, relying on pretextual grounds, had taken a se*148ries of retaliatory actions against respondent because he had testified truthfully on two occasions, one relating to personnel policy and the other involving a public controversy of importance to the Mayor and the members of his cabinet. No matter how narrowly the Court may define the standards for imposing liability upon municipalities in § 1983 litigation, the judgment entered by the District Court in this case should be affirmed.

In order to explain why I believe that affirmance is required by this Court’s precedents,1 it is necessary to begin with a more complete statement of the disputed factual issues that the jury resolved in respondent’s favor, and then to comment on the procedural posture of the case. Finally, I shall discuss the special importance of the character of the wrongful conduct disclosed by this record.

I

The city of St. Louis hired respondent as a licensed architect in 1968. During the ensuing decade, he was repeatedly *149promoted and consistently given “superior” performance ratings. In April 1980, while serving as the Director of Urban Design in the Community Development Agency (CDA), he was recommended for a two-step salary increase by his immediate superior. See 3 Record 1-51.

Thereafter, on two occasions he gave public testimony that was critical of official city policy. In 1980 he testified before the Civil Service Commission (CSC) in support of his successful appeal from a 15-day suspension. In that testimony he explained that he had received advance oral approval of his outside employment and voiced his objections to the requirement of prior written approval.2 The record demonstrates *150that this testimony offended his immediate superiors at the CDA.3

In 1981 respondent testified before the Heritage and Urban Design Commission (HUD) in connection with a pro*151posal to acquire a controversial rusting steel sculpture by Richard Serra. In his testimony he revealed the previously undisclosed fact that an earlier city administration had rejected an offer to acquire the same sculpture, and also explained that the erection of the sculpture would , require the removal of structures on which the city had recently expended about $250,000.4 This testimony offended top offi*152cials of the city government, possibly including the Mayor, who supported the acquisition of the Serra sculpture, as well as respondent’s agency superiors.5 They made it perfectly *153clear that they believed that respondent had violated a duty of loyalty to the Mayor by expressing his personal opinion about the sculpture. Thus, defendant Hamsher testified:

*154“I’m not fond of the sculpture and wasn’t then. But the mayor was elected by the people and he made the decision. He was going to support the installation of the sculpture.

“Therefore, it was my responsibility and the responsibility of others who worked for my agency to do so *155as well and not to express personal opinions in public forums about what that sculpture was going to be and what it would look like.” 5 id., at 3-180.

Defendant Kindleberger made the same point:

“Well, I think the obligation for a senior management individual is to represent fairly the position of his boss which, in our case, happens to be the mayor. And I would — I just think that is something that is appropriate for senior management to do.” 5 id., at 3-250.

After this testimony respondent was the recipient of a series of adverse personnel actions that culminated in his transfer from an important management-level professional position to a rather menial assignment for which he was “grossly over qualified,” 3 id., at 1-80, and his eventual layoff.6 In *156preparing respondent’s service ratings after the Serra sculpture incident, his superiors followed a “highly unusual” procedure that may have violated the city’s personnel regulations.7 Moreover, management officials who were involved in implementing the decision to transfer respondent to a menial assignment made it clear that “there was no reason” for the transfer — except, it would seem, for the possible con*157nection with “the Serra sculpture incident.”8 It is equally clear that the city’s asserted basis for respondent’s ultimate layoff in 1983 — a lack of funds — was pretextual.9

Thus, evidence in the record amply supports the conclusion that respondent was first transferred and then laid off, not for fiscal and administrative reasons, but in retaliation for his public testimony before the CSC and HUD.10 It is undis*158puted that respondent’s right to testify in support of his civil service appeal and his right to testify in opposition to the city’s acquisition of the Serra sculpture were protected by the First Amendment to the Federal Constitution. Given the jury’s verdict, the case is therefore one in which a municipal employee’s federal constitutional rights were violated by officials of the city government. There is, however, a dispute over the identity of the persons who were responsible for that violation. At trial, respondent relied on alternative theories: Either his immediate superiors at CDA (who were named as individual defendants) should be held accountable, or, if the decisions were made at a higher level of government, the city should be held responsible.

The record contains a good deal of evidence of participation in the constitutional tort by respondent’s superiors at CDA, by those directly under the Mayor, and perhaps by the Mayor himself.11 Moreover, in closing argument, defense counsel *159attempted to exonerate the three individual defendants by referring to the actions of higher officials who were not named as defendants.12

*160Thus, we have a case in which, after a full trial, a jury reasonably concluded that top officials in a city’s administration, possibly including the Mayor, acting under color of *161law, took retaliatory action against a gifted but freethinking municipal employee for exercising rights protected by the First Amendment to the Federal Constitution. The legal *162question is whether the city itself is liable for such conduct under § 1983.13

II

In the trial court there was little, if any, dispute over the governing rules of law. In advance of trial, the city filed a *163motion for summary judgment that the District Court ultimately denied because the record contained an affidavit stating that respondent “was transferred due to ‘connivance’ of the mayor, the mayor’s chief of staff, and the city’s personnel director.” 1 Record 130. No one appears to have questioned the proposition that if such facts could be proved at trial, the city could be held liable.14

*164After respondent’s evidence had been presented at trial, the city made a motion for a directed verdict, again advancing the argument that there was insufficient evidence in- the record to support a judgment against the city. The argument on that motion does not indicate that the parties had any dispute about the applicable rules of law. For counsel for the city argued:

“I understand that you can be liable — a municipality can be held liable if its high ranking officials are allowed to violate someone’s constitutional rights. I fail to see how you can find any evidence that the City of St. Louis did that.” 5 id., at 3-28.

The jury obviously disagreed with this assessment of the evidence. Moreover, the judge denied that motion, initially and at the close of all evidence, as well as the city’s motion for a judgment notwithstanding the verdict.

Finally, the ultimate instruction to the jury on the issue of municipal liability was in fact proposed by the city’s attorney, as the plurality acknowledges, ante, at 119; see Brief for Respondent 48; Reply Brief for Petitioner 6:

“As a general principle, a municipality is not liable under 42 U. S. C. § 1983 for the actions of its employees. However, a municipality may be held liable under 42 U. S. C. § 1983 if the allegedly unconstitutional act was *165committed by an official high enough in the government so that his or her actions can be said to represent a government decision.” Instruction No. 15, App. 113.15

In my opinion it is far too late for the city to contend that the jury instructions on municipal liability were insufficient or erroneous.16 In Oklahoma City v. Tuttle, 471 U. S. 808 (1985), we permitted an objection to an instruction by defendant for the first time on appeal only because plaintiff failed to raise the contemporaneous-objection argument until its brief on the merits in this Court. We stated that such arguments “should be brought to our attention no later than in respondent’s brief in opposition to the petition for certiorari.” Id., at 816 (emphasis in original). In this case, respondent properly pointed out in his response to the petition for a writ of *166certiorari that petitioner had failed to object to the relevant jury instruction. Brief in Opposition 10-11.17

Apparently acknowledging that this case cannot be decided on the basis of any possible error in any of the jury instructions, the plurality views petitioner’s motions for summary judgment and a directed verdict as raising and preserving a legal question concerning the standard for determining municipal liability. Ante, at 120. But these motions did not raise any legal issue that was disputed. It is most unfair to permit a defeated litigant in a civil case tried to a verdict before a jury to advance legal arguments that were not made in the District Court, especially when that litigant agrees, both in its motions and proposed instructions, with its opponent’s view of the law.18 Although, as the plurality points out, the *167question presented in the certiorari petition “was manifestly framed in light of the holding of the Court of Appeals,” ante, at 119, the legal issue of municipal liability had never been raised in the District Court.

Given the procedural history, it is not only unfair to respondent, but also poor judicial practice, to use this case as a bulldozer to reshape “a legal landscape whose contours are ‘in a state of evolving definition and uncertainty.’ ” Ante, at 120 (plurality opinion) (citation omitted). It would be far wiser in the long run simply to resolve the issues that have been properly framed by the litigants and preserved for review. Nevertheless, in view of the fact that the Court has “set out again to clarify the issue that we last addressed in Pembaur,” ante, at 124 (plurality opinion), it is appropriate to explain my view of how our precedents in this area apply to this case.

Ill

In Monell v. New York City Dept. of Social Services, 436 U. S. 658 (1978), we held that municipal corporations are “persons” within the meaning of 42 U. S. C. § 1983. Since a corporation is incapable of doing anything except through the agency of human beings, that holding necessarily gave rise to the question of what human activity undertaken by agents of the corporation may create municipal liability in § 1983 litigation.19

The first case dealing with this question was, of course, Monell, in which female employees of the Department of So*168cial Services and the Board of Education of New York City challenged the constitutionality of a citywide policy concerning pregnancy leave. Once it was decided that the city was a “person,” it obviously followed that the city had to assume responsibility for that policy. Even if some departments had followed a lawful policy, I have no doubt that the city would nevertheless have been responsible for the decisions made by either of the two major departments that were directly involved in the litigation.

In Owen v. City of Independence, 445 U. S. 622 (1980), the Court held that municipalities are not entitled to qualified immunity based on the good faith of their officials. As a premise to this decision, we agreed with the Court of Appeals that the city “was responsible for the deprivation of petitioner’s constitutional rights.” Id., at 633; see also id., at 655, n. 39. Petitioner had been fired as City Chief of Police without a notice of reasons and without a hearing, after the City Council and the City Manager had publicly reprimanded him for his administration of the Police Department property room. This isolated personnel action was clearly not taken pursuant to a rule of general applicability; nonetheless, we had no problem with the .Court of Appeals’ conclusion that the action of the City Council and City Manager was binding on the city.20

*169In the next municipal liability case, the Court held that an isolated unconstitutional seizure by a sole police officer did not bind the municipality. Oklahoma City v. Tuttle, 471 U. S. 808 (1985).21 Thus, that holding rejected the common-law doctrine of respondeat superior as the standard for measuring municipal liability under § 1983. It did not, of course, reject the possibility that liability might be predicated on the conduct of management level personnel with policymaking authority.

Finally, in Pembaur v. Cincinnati, 475 U. S., at 471, we definitively held that a “decision by municipal policymakers on a single occasion” was sufficient to support an award of damages against the municipality. In Pembaur, a County Prosecutor had advised County Sheriffs at the doorstep of a recalcitrant doctor to “go in and get [the witnesses]” to alleged charges of fraud by the doctor. Id., at 473. Because the Sheriffs possessed only arrest warrants for the witnesses and not a search warrant for the doctor’s office as well, the *170advice was unconstitutional, see Steagald v. United States, 451 U. S. 204 (1981), and the question was whether the County Prosecutor’s isolated act could subject the county to damages under § 1983 in a suit by the doctor. In the part of his opinion that commanded a majority of the Court, Justice Brennan wrote:

“[A] government frequently chooses a course of action tailored to a particular situation and not intended to control decisions in later situations. If the decision to adopt that particular course of action is properly made by that government’s authorized decisionmakers, it surely represents an act of official government ‘policy’ as that term is commonly understood. More importantly, where action is directed by those who establish governmental policy, the municipality is equally responsible whether that action is to be taken only once or to be taken repeatedly.” Pembaur v. Cincinnati, 475 U. S., at 481 (footnote omitted).

Since the County Prosecutor was authorized to establish law enforcement policy, his decision in that area could be attributed to the county for purposes of § 1983 liability. As Justice Powell correctly pointed out in his dissent, “the Court . . . focus[ed] almost exclusively on the status of the decision-maker.” Id., at 498.

Thus, the Court has permitted a municipality to be held liable for the unconstitutional actions of its agents when those agents enforced a rule of general applicability (Monell); were of sufficiently high stature and acted through a formal process (Owen); or were authorized to establish policy in the particular area of city government in which the tort was committed (Pembaur). Under these precedents, the city of St. Louis should be held liable in this case.

Both Pembaur and the plurality and concurring opinions today acknowledge that a high official who has ultimate control over a certain area of city government can bind the city *171through his unconstitutional actions even though those actions are not in the form of formal rules or regulations. See Pembaur v. Cincinnati, supra, at 479-481; ante, at 123 (plurality), at 139-140 (concurrence). Although the Court has explained its holdings by reference to the nonstatutory term “policy,” it plainly has not embraced the standard understanding of that word as covering a rule of general applicability. Instead it has used that term to include isolated acts not intended to be binding over a class of situations. But when one remembers that the real question in cases such as this is not “what constitutes city policy?” but rather “when should a city be liable for the acts of its agents?”, the inclusion of single acts by high officials makes sense, for those acts bind a municipality in a way that the misdeeds of low officials do not.

Every act of a high official constitutes a kind of “statement” about how similar decisions will be carried out; the assumption is that the same decision would have been made, and would again be made, across a class of cases. Lower officials do not control others in the same way. Since their actions do not dictate the responses of various subordinates, those actions lack the potential of controlling governmental decisionmaking; they are not perceived as the actions of the city itself. If a county police officer had broken down Dr. Pembaur’s door on the officer’s own initiative, this would have been seen as the action of an overanxious officer, and would not have sent a message to other officers that similar actions would be countenanced. One reason for this is that the County Prosecutor himself could step forward and say “that was wrong”; when the County Prosecutor authorized the action himself, only a self-correction would accomplish the same task, and until such time his action would have countywide ramifications. Here, the Mayor, those working for him, and the agency heads are high-ranking officials; accordingly, we must assume that their actions have citywide ramifications, both through their similar response to a like *172class of situations, and through the response of subordinates who follow their lead.22

Just as the actions of high-ranking and low-ranking municipal employees differ in nature, so do constitutional torts differ. An illegal search (Pembaur) or seizure (Tuttle) is quite different from a firing without due process (Owen); the retaliatory personnel action involved in today’s case is in still another category. One thing that the torts in Pembaur, Tuttle, and Owen had in common is that they occurred “in the open”; in each of those cases, the ultimate judgment of unconstitutionality was based on whether undisputed events (the breaking-in in Pembaur, the shooting in Tuttle, the firing in Owen) comported with accepted constitutional norms. But *173the typical retaliatory personnel action claim pits one story against another; although everyone admits that the transfer and discharge of respondent occurred, there is sharp, and ultimately central, dispute over the reasons — the motivation— behind the actions. The very nature of the tort is to avoid a formal process. Owen’s relevance should thus be clear. For if the Court is willing to recognize the existence of municipal policy in a nonrule case as long as high enough officials engaged in a formal enough process, it should not deny the existence of such a policy merely because those same officials act “underground,” as it were. It would be a truly remarkable doctrine for this Court to recognize municipal liability in an employee discharge case when high officials are foolish enough to act through a “formal process,” but not when similarly high officials attempt to avoid liability by acting on the pretext of budgetary concerns, which is what this jury found based on the evidence presented at trial.

Thus, holding St. Louis liable in this case is supported by both Pembaur and Owen. We hold a municipality liable for the decisions of its high officials in large part because those decisions, by definition, would be applied across a class of cases. Just as we assume in Pembaur that the County Prosecutor (or his subordinates) would issue the same breakdown-the-door order in similar cases, and just as we assume in Owen that the City Council (or those following its lead) would fire an employee without notice of reasons or opportunity to be heard in similar cases, so too must we assume that whistleblowers like respondent would be dealt with in similar retaliatory fashion if they offend the Mayor, his staff, and relevant agency heads, or if they offend those lower ranking officials who follow the example of their superiors. Furthermore, just as we hold a municipality liable for discharging an employee without due process when its city council acts formally — for a due process violation is precisely the type of constitutional tort that a city council might commit when it acts formally — so too must we hold a municipality liable for discharging an employee in retaliation against his public speech *174when similarly high officials act informally — for a First Amendment retaliation tort is precisely the type of constitutional tort that high officials might commit when they act in concert and informally.

Whatever difficulties the Court may have with binding municipalities on the basis of the unconstitutional conduct of individuals, it should have no such difficulties binding a city when many of its high officials — including officials directly under the Mayor, agency heads, and possibly the Mayor himself — cooperate to retaliate against a whistleblower for the exercise of his First Amendment rights.23

I would affirm the judgment of the Court of Appeals.

TJOFLAT, Circuit Judge:

I.

Michael Holloman, a former student at Parrish High School in Walker County, Alabama, filed a § 1983 suit against Fawn Allred, his economics and government teacher; George Harland, the school principal; and the Walker County Board of Education (“School Board”), which oversaw the school. He claimed that his rights under the First Amendment’s Speech Clause were violated when Allred and Harland punished him for silently raising his fist during the daily flag salute instead of reciting the Pledge of Allegiance with the rest of his class. He further claims *1260that his Establishment Clause rights were violated by Allred’s daily “ritual” of conducting a silent moment of prayer. He sought both legal and equitable relief.

The district court granted summary judgment on both claims to Allred and Harland on qualified immunity grounds. In a separate opinion, it granted summary judgment to the School Board, concluding that Holloman failed to articulate a violation of his constitutional rights or demonstrate a way in which the Board (as a municipal governing entity) could be held liable for the acts at issue here. Holloman appeals both rulings.

Subpart A of this Part examines the facts supporting Holloman’s Speech Clause claim. Subpart B explains his Establishment Clause allegations. Subpart C sets forth the framework of state statutes and School Board regulations implicated by Holloman’s claims, and Subpart D delves into the procedural history of this case in greater detail. Throughout this discussion, because we are reviewing grants of summary judgment to the defendants, we view the evidence in the light most favorable to the plaintiff. See Johnson v. Governor of Florida, 353 F.3d 1287, 1292 (11th Cir.2003).

A.

Holloman contends that Allred and Har-land violated his First Amendment right to free speech (as incorporated against the states through the Fourteenth Amendment’s Due Process Clause) by treating him adversely because he silently raised his fist during the flag salute instead of reciting the Pledge of Allegiance.1 To understand what happened, it is necessary to consider their treatment of another student, John Michael Hutto, the day before their confrontation with Holloman.

1.

Allred taught her Economics and Government class in the first period of each day, during which time the Pledge of Allegiance was recited over the school intercom system. It was customary for students to stand by their desks, with their hands over their hearts, and recite the pledge.

During the flag salute on May 16, 2000, Hutto remained silent with his hands in his pockets, without causing a disturbance. When Allred asked him why he was not participating in the flag salute, Hutto responded that he “didn’t want to say it, he didn’t have to say it, and he hadn’t said it for a month.” Allred stated, “You don’t want to say the pledge and the United States Air Force Academy has given you a scholarship?,” then continued class.

At lunch that day, Allred told Harland of Hutto’s refusal to say the pledge. Har-land became very angry and met with Allred, Hutto, and Vice Principal Jason Adkins in his office. Harland told Hutto that he was disappointed in Hutto’s refusal to salute the flag, and threatened to report the incident to both Hutto’s recruiter at the Air Force Academy as well as the Congressman who had recommended Hut-to to the Academy. Harland also ordered Hutto to apologize to Allred and her class for refusing to salute the flag.

Later that day, Harland went to Hutto’s physics class (in which Holloman was a student) and declared that “anyone who joined in [Hutto’s] protest and refused to say the pledge or committed similar action would be punished.” The following day, Hutto recited the Pledge of Allegiance with the rest of the class, and the day after that he apologized to Allred and her students.

*12612.

During the flag salute on the day after the Hutto incident, Holloman stood with the other students in Allred’s.class, but did not recite the Pledge of Allegiance. Instead, he silently raised his fist in the air while the rest of the class recited the pledge; once the pledge was over, he sat down like everyone else. He did not say anything, touch any other students, disrupt the class, or obstruct anyone’s view of the flag. Allred, however, immediately chastised him in front of the class, saying that he had acted inappropriately and “disrespectfully],” and that she was “disappointed.” She then started class in her normal fashion.

Later that day, Allred informed Harland of what happened, and Harland summoned Allred and Holloman into the principal’s office. Holloman explained that he had raised his fist “in protest of what happened to [Hutto].” Harland told Holloman “how disappointed he was, and that he felt that he had failed teaching Michael Holloman responsibility, morals and values.” He also informed Holloman that he would have to serve three days’ detention and could not receive his diploma until after he completed his punishment. In addition, Harland required Holloman to apologize to Allred’s class. When Holloman left Har-land’s office, Harland called Holloman’s mother, explaining “that he was too mad and upset to punish Michael at the time because he may hurt Michael.”

Since graduation was that Friday, there was not enough time left in the school year for Holloman to serve his detentions while still being able to receive his diploma on graduation day. Harland consequently offered Holloman the opportunity to receive a paddling instead. Holloman agreed and, with Allred watching, was paddled by Har-land.

B.

Allred began her Economics and Government class almost every day by asking, “Does anyone have any prayer requests?” After her students offered various dedications, Allred would hold a moment of silence. Allred frequently opened this moment of silence by saying “Let us pray,” and often ended it by saying “Amen.” All-red explicitly states that over the 1999-2000 school year, this practice became a daily “ritual.” She never told her students that they were free to leave the room during either her prayer requests or the subsequent moment of silent prayer.

One day, Vice Principal Adkins sat in on her class and personally observed this phenomenon. When Allred attempted to begin her economics lesson, one of her students raised her hand and reminded Allred that she had forgotten to elicit her customary prayer requests. At that point, Allred took prayer requests from the class, then commenced a moment of silence by saying, “Let us pray.” On another occasion, at the conclusion of the moment of silence, Allred permitted one of her students to read aloud a passage from the Bible.

C.

The events in this case did not occur in a vacuum. In 1995, the Alabama state legislature enacted a statute which required the State Board of Education and all local school boards to

develop and implement ... a comprehensive character education program for all grades to consist of not less than ten minutes of instruction per day focusing upon the students’ development of the following character traits: courage, patriotism, citizenship, honesty, fairness, respect for others, kindness, cooperation, self-respect, self-control, courtesy, compassion, tolerance, dili*1262gence, generosity, punctuality, cleanliness, cheerfulness, school pride, respect for the environment, patience, creativity, sportsmanship, loyalty, and perseverance. Each plan of instruction shall include the Pledge of Allegiance to the American flag.

Ala.Code § 16-6B-2(h). This law made daily recitation of the Pledge of Allegiance a part of the character education program the Legislature required local school boards to implement. A separate statute, however, emphasized that students should not be forced to recite the pledge. See Aa.Code § 16-43-5 (“The State Board of Education shall afford all students attending public kindergarten, primary and secondary schools the opportunity each school day to voluntarily recite the pledge of allegiance to the United States flag.” (emphasis added)).

To implement these requirements, Larry Banks, the Superintendent of the Walker County School District, sent a letter on behalf of the Walker County Board of Education to all the principals in the district, stating, “[EJach school system must incorporate a Character Education Plan which will consist of 10 minutes of instruction per day in various areas, such as, the Pledge of Allegiance.... Each day must include the Pledge of Allegiance and then other areas as mentioned as you determine at your school.”

Banks also sent each principal a form to complete to specify how each school intended to incorporate into its curriculum the character-education requirements set forth above. The memo directed, “Please begin to make plans and be prepared to submit your local Character Education Plan [to the county school board] which must be forwarded to the State Superintendent’s Office.” The County School Board apparently had to either review or approve each school’s character education plan before it was forwarded to the State. Aired contends that her daily moment of silent prayer was conducted in partial fulfillment of these character education requirements — it was intended to teach compassion.

D.

On July 3, 2000, Holloman and Hutto filed a class action suit under 42 U.S.C. § 1983 in the Northern District of Alabama against the Walker County Board of Education, Harland, Adkins, and Aired. They alleged that their First Amendment rights had been violated because they had been chastised, threatened, and punished for refusing to say the Pledge of Ale-giance. They also claimed that Alred’s practice of soliciting prayer requests and setting aside a moment of silence for prayer violated the Establishment Clause. The complaint sought compensatory and punitive damages, as well as declaratory and injunctive relief.

A few months later, an amended complaint was filed; it was substantially identical to the original except Hutto was no longer a party. The district court dismissed Jason Adkins as a defendant (with Holloman’s consent), and declined to certify the class action. Holloman does not appeal either of these rulings.

In their answer to Holloman’s amended complaint, Aired and Harland cited qualified immunity as an affirmative defense. They later moved the court for summary judgment on qualified immunity grounds. The court granted their motion and dismissed them from the case. It concluded that Holloman’s Speech Clause allegations did not constitute a First Amendment violation, and certainly not a violation of a right that was “clearly established” at the time of the incidents. The court also rejected Holloman’s Establishment Clause claims, stating there was no Eleventh Cir*1263cuit precedent during the 1999-2000 school year that “clearly established” a moment of silence for prayer as being unconstitutional. Holloman appealed the court’s decision to grant Allred and Harland summary judgment on qualified immunity grounds.2

Following the district court’s dismissal of Holloman’s claims against Allred and Harland, the Board of Education also sought summary judgment. The court granted the Board’s motion, concluding that Holloman had not alleged facts sufficient to hold the Board liable under § 1983. Holloman appeals this ruling as well, arguing that he stated valid claims against the School Board.3

In these consolidated appeals, we review the district court orders granting the defendants summary judgment. Part II of this opinion explains why the district court erred in granting Allred and Harland summary judgment on qualified immunity grounds against Holloman’s Speech Clause claims. Part III shows that Allred is not even potentially entitled to summary judgment on qualified immunity grounds against Holloman’s Establishment Clause claims because she has not established as a matter of law that, in holding her daily moment of silent prayer, she was engaged in a discretionary function of her job. Part IV assesses Holloman’s underlying Establishment Clause claim, concluding that he has introduced evidence sufficient to support a determination that a clearly established right has been violated.4 Part V discusses how Holloman has successfully articulated several theories under which the School Board may be held liable for the Speech and Establishment Clause violations, while Part VI briefly concludes.

II.

Section 1988 provides:

Every person who, under color of any statute, ordinance, regulation, custom, or usage of any State ... subjects, or causes to be subjected, any citizen of the United States ... to the deprivation of any rights, privileges, or immunities secured by the Constitution and laws, shall be liable to the party injured in an action at law [or] suit in equity....

42 U.S.C. § 1983. There are two ways in which an individual may be held liable under § 1983 — he may be sued for his own personal actions (“direct liability”), or, under certain limited circumstances, for the actions of his subordinates (“supervisoral liability”), see, e.g., Lewis v. Smith, 855 F.2d 736, 738 (11th Cir.1988).5

When a government official is sued under a theory of direct liability, he may seek summary judgment on qualified immunity grounds.6 To even be potential*1264ly eligible for summary judgment due to qualified immunity, the official must have been engaged in a “discretionary function” when he performed the acts of which the plaintiff complains. Harlow v. Fitzgerald, 457 U.S. 800, 818, 102 S.Ct. 2727, 2738, 73 L.Ed.2d 396 (1982) (holding that qualified immunity extends to “government officials performing discretionary functions”). It is the burden of the governmental official to make this showing. Storck v. City of Coral Springs, 354 F.3d 1307, 1314 (2003) (“Under qualified immunity analysis, the public official must first prove that he was acting within the scope of his discretionary authority when the allegedly unconstitutional acts took place.” (emphasis added)). A defendant unable to meet this burden may not receive summary judgment on qualified immunity grounds. Lumley v. City of Dade City, 327 F.3d 1186, 1194 (11th Cir.2003) (“If the defendants were not acting within their discretionary authority, they are ineligible for the benefit of qualified immunity.”); see also Lee v. Ferraro, 284 F.3d 1188, 1194 (11th Cir.2002). While a number of our cases omit this step of the analysis, see, e.g., Denno v. Sch. Bd., 218 F.3d 1267 (11th Cir.2000); Hall v. Talladega City Bd. of Educ., 115 F.3d 821 (11th Cir.1997), binding Supreme Court and Eleventh Circuit precedents require us to consider expressly this critical threshold matter. We explain this “discretionary function” test in greater detail in Subpart II.A.

If, interpreting the evidence in the light most favorable to the plaintiff, the court concludes that the defendant was engaged in a discretionary function, then the burden shifts to the plaintiff to show that the defendant is not entitled to qualified immunity. Cottone v. Jenne, 326 F.3d 1352, 1358 (11th Cir.2003) (“Once a defendant establishes that he was acting within his discretionary authority, the burden shifts to the plaintiff to show that the defendant is not entitled to qualified immunity”). To overcome qualified immunity, the plaintiff must satisfy a two prong test; he must show that: (1) the defendant violated a constitutional right, and (2) this right was clearly established at the time of the alleged violation.7 Wilson v. Layne, 526 U.S. 603, 609, 119 S.Ct. 1692, 1697, 143 L.Ed.2d 818 (1999) (“A court evaluating a claim of qualified immunity must first determine whether the plaintiff has alleged the deprivation of an actual constitutional right at all, and if so, proceed to determine whether that right was clearly established at the time of the alleged violation.”) (quotations and citations omitted). If the plaintiff prevails on both prongs of this test, then the defendant is unable to obtain summary judgment on qualified immunity grounds.

Applying this test, the district court awarded Allred and Harland summary judgment against Holloman’s Speech Clause claims. The Speech Clause of the First Amendment protects at least two separate, yet related, rights: (1) the right to freedom of expression, and (2) the right to be free from compelled expression. United States v. United Foods, Inc., 533 U.S. 405, 410, 121 S.Ct. 2334, 2338, 150 L.Ed.2d 438 (2001). These rights unquestionably exist in public schools. See Tink*1265er v. Des Moines Indep. Comm. Seh. Dist., 393 U.S. 503, 506, 89 S.Ct. 733, 736, 21 L.Ed.2d 731 (1969).

Holloman argues that Allred and Har-land are directly liable for violating both constitutional rights guaranteed by the Speech Clause. First, he maintains, they violated his right to be free from compelled expression by chastising, threatening, and ultimately punishing him for failing to recite the Pledge of Allegiance. Second, to the extent that he was punished for silently raising his fist in the air during the Pledge of Allegiance (rather than for simply failing to recite the pledge), Hollo-man contends that Allred and Harland violated his right to engage in affirmative expression. Finally, regardless of whether Holloman’s expression was constitutionally protected in itself, he has the First Amendment right to be free of viewpoint-based discrimination and punishment.

We conclude that neither Allred nor Harland are not entitled to summary judgment on qualified immunity grounds against any of these First Amendment claims. In Subpart II.A, we conclude that both Allred and Harland were engaged in a discretionary function at the time they disciplined Holloman in connection with the flag salute incident, and so are potentially entitled to summary judgment on qualified immunity grounds. In Subpart II.B, we discuss how the evidence — interpreted in the light most favorable to Hollo-man — supports the conclusion that his clearly-established right to be free from compelled speech was violated. Subpart II.C makes a similar finding regarding his right to engage in affirmative expression. Subpart II.D concludes that, even if — as the dissent contends — Holloman’s expression was not itself constitutionally protected, Allred’s behavior nevertheless violated the First Amendment because she punished him for expressing a viewpoint she found repugnant, rather than for any disruption he purportedly caused (which, interpreting the evidence in his favor, was entirely negligible). Consequently, Hollo-man has made the necessary showings with regard to his three Speech Clause claims to overcome Allred’s and Harland’s assertions of qualified immunity at this stage.

A.

In many areas other than qualified immunity, a “discretionary function” is defined as an activity requiring the exercise of independent judgment, and is the opposite of a “ministerial task.” See, e.g., Williams v. Wood, 612 F.2d 982, 985 (5th Cir.1980).8 In the qualified immunity context, however, we appear to have abandoned this “discretionary function/ministerial task” dichotomy. In McCoy v. Webster, 47 F.3d 404, 407 (11th Cir.1995), we interpreted “the term ‘discretionary authority’ to include actions that do not necessarily involve an element of choice,” and emphasized that, for purposes of qualified immunity, a governmental actor engaged in purely ministerial activities can nevertheless be performing a discretionary function.

Instead of focusing on whether the acts in question involved the exercise of actual discretion, we assess whether they are of a type that fell within the employee’s job responsibilities. Our inquiry is two-fold. We ask whether the government employee was (a) performing a legitimate job-related function (that is, pursuing a job-related goal), (b) through means that were within his power to utilize. See *1266Hill v. Dekalb Reg’l Youth Det. Ctr., 40 F.3d 1176, 1185 n. 17 (11th Cir.1994) (“A government official acts within his or her discretionary authority if objective circumstances compel the conclusion that challenged actions occurred in the performance of the official’s duties and within the scope of this authority.” (emphasis added)).

One might reasonably believe that violating someone’s constitutional rights is never a legitimate job-related function or within the scope of a government official’s authority or power. As we explained in Harbert Int'l Inc. v. James, 157 F.3d 1271, 1282 (11th Cir.1998) (quotation marks and citation omitted), however, “the inquiry is not whether it was within the defendant’s authority to commit the allegedly illegal act. Framed that way, the inquiry is no more than an untenable tautology.” In applying each prong of this test, we look to the general nature of the defendant’s action, temporarily putting aside the fact that it may have been committed for an unconstitutional purpose, in an unconstitutional manner, to an unconstitutional extent, or under constitutionally inappropriate circumstances.

Consider the first prong of the test- — -whether the official is engaged in a legitimate job-related function. In Sims v. Metropolitan Dade County, 972 F.2d 1230 (11th Cir.1992), “we did not ask whether it was within the defendant’s authority to suspend an employee for an improper reason; instead, we asked whether [the defendant’s] discretionary duties included the administration of discipline.” Harbert, 157 F.3d at 1282. Similarly, in assessing whether a police officer may assert qualified immunity against a Fourth Amendment claim, we do not ask whether he has the right to engage in unconstitutional searches and seizures, but whether engaging in searches and seizures in general is a part of his job-related powers and responsibilities. See, e.g., Madiwale v. Savaiko, 117 F.3d 1321, 1324 (11th Cir.1997). Put another way, to pass the first step of the discretionary function test for qualified immunity, the defendant must have been performing a function that, but for the alleged constitutional infirmity, would have fallen with his legitimate job description.

Of course, we must be sure not to characterize and assess the defendant’s act at too high a level of generality. Nearly every act performed by a government employee can be described, in general terms, as ostensibly “furthering the public interest.” If we jump to such a high level of abstraction, it becomes impossible to determine whether the employee was truly acting within the proper scope of his job-related activities. Consequently, we consider a government official’s actions at the minimum level of generality necessary to remove the constitutional taint. In considering whether an act of allegedly excessive force fell within a police officer’s duties, for example, we do not ask whether police have the right to use excessive force. We also do not immediately jump to a high level of generality and ask whether police are responsible for enforcing the law or promoting the public interest. We instead ask whether they have the power to attempt to effectuate arrests. See, e.g., Ferraro, 284 F.3d at 1194 (holding, in an excessive force suit, “there can be no doubt that [the police officer defendant] was acting in his discretionary capacity when he arrested [plaintiff]”).

After detérmining that an official is engaged in a legitimate job-related function, it is then -necessary to turn to the second prong of the test and determine whether he is executing that job-related function — that is, pursuing his job-related goals — in an authorized manner. The primary purpose of the qualified immunity *1267doctrine is to allow government employees to enjoy a degree of protection only when exercising powers that legitimately form a part of their jobs. See, e.g., Harlow, 457 U.S. at 819 & n. 34, 102 S.Ct. at 2739 & n. 34 (limiting the availability of qualified immunity to situations where “an official’s duties legitimately require action” and to “actions within the scope of an official’s duties”). Each government employee is given only a certain “arsenal” of powers with which to accomplish her goals. For example, it is not within a teacher’s official powers to sign her students up for the Army to promote patriotism or civic virtue, or to compel them to bring their property to school to redistribute their wealth to the poor so that they can have firsthand experience with altruism.

Employment by a local, county, state, or federal government is not a carte blanche invitation to push the envelope and tackle matters far beyond one’s job description or achieve one’s official goals through unauthorized means. Pursuing a job-related goal through means that fall outside the range of discretion that comes with an employee’s job is not protected by qualified immunity.

Under this standard, Allred—as a matter of law—was undoubtedly engaged in a discretionary function in chastising Holloman for raising his fist during the Pledge of Allegiance and later referring him to Harland for punishment. Though Allred is not empowered to violate constitutional rights as part of her official duties, she did have the responsibility of maintaining decorum in the classroom. The fact that she may have attempted to keep order in the classroom in an unconstitutional manner does not change the fact that she was fulfilling a legitimate job-related function. Moreover, the ways in which she attempted to pursue this job-related goal (chastising Holloman and reporting him to the principal)—examined on a general level rather than in this specific application— were legitimate prerogatives of her job. From an alternate perspective, putting aside Holloman’s First Amendment claim, Allred’s actions would undoubtedly be considered part of her duties and legitimate exercises of her authority. Consequently, under the two-prong test articulated above, her activities in relation to the flag salute incident were discretionary acts for which she may seek qualified immunity.

For similar reasons, Harland is also potentially entitled to qualified immunity against Holloman’s Speech Clause claims. Disciplining students is a legitimate discretionary function performed by principals. See Kirkland v. Greene County Bd. of Educ., 347 F.3d 903, 903 n. 1 (11th Cir.2003), and in the State of Alabama, spanking students is a legitimate part of a principal’s “arsenal” for enforcing such discipline. Consequently, the burden shifts to Holloman to demonstrate that Allred and Harland are not entitled to summary judgment on qualified immunity grounds.

B.

As discussed earlier, once a defendant establishes that he was engaged in a discretionary function at the time of the acts in question, the burden shifts to the plaintiff to show that the defendant is not entitled to summary judgment on qualified immunity grounds. To do so, the plaintiff must demonstrate that a reasonable jury could interpret the evidence in the record as showing that the defendant violated a constitutional right that was clearly established at the time of the acts in question.

We begin by examining Holloman’s claim that Allred and Harland violated his First Amendment right to be free from compelled speech. Section 1 considers *1268whether Holloman has successfully articulated a violation of a constitutional right, while Section 2 analyzes whether such a right was clearly established at the time of the incidents. Based on these discussions, we conclude that, interpreting the evidence in the light most favorable to Holloman, both Allred and Harland engaged in acts amounting to violations of Holloman’s right to be free from compelled speech. Consequently, neither defendant is entitled to summary judgment on qualified immunity grounds against this Speech Clause claim.

1.

The Speech Clause of the First Amendment states, “Congress shall make no law ... abridging the freedom of speech.... ” U.S. Const, amend. I. The First Amendment, as incorporated through the Due Process Clause of the Fourteenth Amendment, Near v. Minnesota, 283 U.S. 697, 707, 51 S.Ct. 625, 628, 75 L.Ed. 1357 (1931), applies to state and municipal governments, state-created entities, and state and municipal employees, West Virginia State Bd. of Educ. v. Barnette, 319 U.S. 624, 637, 63 S.Ct. 1178, 1185, 87 L.Ed. 1628 (1943).

In Barnette, the Court held that the right to be free from compelled speech protects public school students from being forced to participate in the flag salute. It stated, “[T]he action of the local authorities in compelling the flag salute and pledge transcends constitutional limitations on their power and invades the sphere of intellect and spirit which it is the purpose of the First Amendment to our Constitution to reserve from all official control.” Id. at 642, 63 S.Ct. at 1187.

Several pieces of evidence in the record support Holloman’s contention that he was disciplined for failing to recite the Pledge of Allegiance. First, the day before the Holloman incident, John Michael Hutto was chastised in front of the class, sent to the Principal’s office, threatened that his recommendation to the Air Force Academy would be revoked, and forced to apologize to his teacher and classmates, simply because he remained silent during the Pledge of Allegiance (without engaging in any affirmatively expressive activity). Second, Allred’s deposition is replete with references to her “patriotism” and desire to see the American flag saluted in the “proper” or “normal” way; she was deeply offended by the notion of Americans not wanting to salute the flag.

Third, according to Holloman’s affidavit, Harland interrupted Holloman’s physics class to explicitly threaten that any students who refused to say the Pledge of Allegiance would be punished. Fourth, the affidavits of both Holloman and his mother state that he was told he was being punished for failing to salute the flag. Indeed, Harland told Holloman’s mother during their phone conversation that he had to wait before disciplining Holloman because he was afraid that he (Harland) would hurt him. Consequently, there is more than enough evidence in the record to allow a reasonable jury to adopt this interpretation of events.

Allred’s acts (if proven at trial), as a matter of law, violated Holloman’s constitutional rights. First, according to All-red’s own testimony, she instructed him that there were only two “permissible” ways of saying the pledge, and that any other way of doing so was prohibited. Second, she verbally chastised him in front of the class for his constitutionally protected actions (either failing to salute the flag or expressing his opinion in a non-disruptive fashion).

Verbal censure is a form of punishment, albeit a mild one. The intent behind this act was to dissuade him from exercising a *1269constitutional right. She singled out Hol-loman in front of his entire class, subjecting him to embarrassment and humiliation. Given the gross disparity in power between a teacher and a student, such comments — particularly in front of the student’s peers — coming from an authority figure with tremendous discretionary authority, whose words carry a presumption of legitimacy, cannot help but have a tremendous chilling effect on the exercise of First Amendment rights. See Riley v. Natl Fed’n of the Blind, 487 U.S. 781, 794, 108 S.Ct. 2667, 2676, 101 L.Ed.2d 669 (1988) (invaliding a “scheme” that “must necessarily chill speech in direct contravention of the First Amendment’s dictates”); Dickerson v. United States, 580 U.S. 428, 459, 120 S.Ct. 2326, 2344, 147 L.Ed.2d 405 (2000) (Scalia, J., dissenting) (“[T]he Court has viewed the importation of ‘chill’ as itself a violation of the First Amendment.”) (emphasis in original). As' in the Establishment Clause context, such “public pressure ... can be as real as any overt compulsion,” particularly in a “classroom setting, where ... the risk of compulsion .is especially high”; such measures may not be used to deter, even if “subtlfy] or indirect[ly],” the exercise of constitutional rights. Lee v. Weisman, 505 U.S. 577, 593, 596, 112 S.Ct. 2649, 2658, 2660, 120 L.Ed.2d 467 (1992).

Finally, Allred played a major role in the administration of Holloman’s more formal punishment, his paddling. She reported the incident to the principal with the intent and hope that Holloman be disciplined, was present during Holloman’s questioning by Harland, did not object or attempt to dissuade Harland in any way, was present for the paddling, and manifested her approval of it throughout the entire process. For similar reasons, Hol-loman has also adduced sufficient evidence to support his claim against Harland, the one who actually spanked him.

Having demonstrated that the record amply supports Holloman’s contention that the defendants violated his constitutional right to be free from compelled speech, we now consider whether this right was clearly established at the time of the events in question.

2.

Barnette, 319 U.S. at 642, 63 S.Ct. at 1187, clearly and specifically established that schoolchildren have the right to refuse to say the Pledge of Allegiance. Under Barnette, any “reasonable person would have known” that disciplining Holloman for refusing to recite the pledge impermissibly chills his First Amendment rights. Thomas v. Roberts, 261 F.3d 1160, 1170 (11th Cir.2001) (citation omitted). Consequently, on these alleged facts, Allred and Harland are not entitled to qualified immunity against Holloman’s claims related to compelled speech. We reverse the district court’s holding to the contrary.

Having assessed the viability of Hollo-man’s First Amendment claim to be free from compelled speech, we now turn to Holloman’s First Amendment claim to engage in affirmative expression.

C.

One of Holloman’s alternate bases for recovery under the First Amendment is that the defendants punished him for engaging in constitutionally protected speech. He maintains, in other words, that even if the defendants punished him for silently raising his fist during the Pledge of Allegiance — rather than for merely remaining silent during the pledge — his rights under the Speech Clause were still violated. The district court dedicated the overwhelming majority of its opinion regarding Allred and Harland to this argument, concluding that *1270“[n]o jury could reasonably conclude that ‘every like-situated, reasonable’ teacher or principal would necessarily know that punishing Holloman for his unorthodox and deliberately provocative and disruptive gesture violated federal law....” (Mem. Op., June 4, 2001, at 10).

We are again fofced to reverse. Section 1 shows that the evidence (interpreted in Holloman’s favor) demonstrates the existence of a limited constitutional right to engage in non-disruptive expression in a classroom environment, while Section 2 demonstrates that this right was clearly established when the defendants chastised and punished him. Consequently, Allred and Harland are not entitled to summary judgment on qualified immunity grounds against Holloman’s Speech Clause claim regarding his right to express himself.

1.

The Constitution guarantees students (and all people) the right to engage not only in “pure speech,” but “expressive conduct,” as well. See United States v. O’Brien, 391 U.S. 367, 376-77, 88 S.Ct. 1673, 1678-79, 20 L.Ed.2d 672 (1968). In Spence v. Washington, 418 U.S. 405, 94 S.Ct. 2727, 41 L.Ed.2d 842 (1974), the Supreme Court held that, to determine whether a particular act counts as expressive conduct, a court must determine whether “[a]n intent to convey a particularized message was present, and in the surrounding circumstances the likelihood was great that the message would be understood by those who viewed it.” Id. at 410-11, 94 S.Ct. at 2730. The Court later liberalized this test, however, emphasizing that “a narrow, succinctly articulable message is not a condition of constitutional protection, which if confined to expressions conveying a ‘particularized message,’ would never reach the unquestionably shielded painting of Jackson Pollock, music of Arnold Schonberg, or Jabberwocky verse of Lewis Carroll.” Hurley v. Irish-Am., Gay, Lesbian & Bisexual Group of Boston, etc., 515 U.S. 557, 569, 115 S.Ct. 2338, 2345, 132 L.Ed.2d 487 (1995). Thus, in determining whether conduct is expressive, we ask whether the reasonable person would interpret it as some sort of message, not whether an observer would necessarily infer a specific message.

At the very least, Holloman’s gesture was expressive conduct. It is quite reasonable to infer that at least some students would have recognized his act for what it was — a protest over Allred’s treatment of Hutto. Even if students were not aware of the specific message Holloman was attempting to convey, his fist clearly expressed a generalized message of disagreement or protest directed toward Allred, the school, or the country in general.

It is quite possible, however, that Hollo-man’s act constituted “pure speech.” As the Court suggested in O’Brien, 391 U.S. at 376, 88 S.Ct. at 1678, expressive conduct is an act with significant “ ‘non-speech’ elements,” that is being used in a particular situation to convey a message. Holloman’s act does not contain any of the substantive “non-speech” elements that are necessary to remove something from the realm of “pure speech” into the realm of expressive conduct. It seems as purely communicative as a sign-language gesture or the act of holding up a sign, and in this respect is similar to the wearing of a black armband, which the Tinker Court found to be a “primary First Amendment right[ ] akin to ‘pure speech.’ ” 393 U.S. at 508, 89 S.Ct. at 737.

It does not ultimately matter whether Holloman’s act is characterized as “pure speech” or “expressive conduct” because this circuit appears to apply the same test in assessing school restrictions on either kind of expression. This Section *1271summarizes these principles and applies them to the instant case.

a.

As with all rights, the scope of the First Amendment has boundaries.- On many occasions, we have affirmed the right of public educational institutions “to adopt and enforce reasonable, non-discriminatory regulations as to the time, place and manner of student expressions and demonstrations.” Bayless v. Martine, 430 F.2d 873, 878 (5th Cir.1970). This “reasonableness” test is not the anemic simulacrum of a constraint on governmental power found in the Due Process Clause’s “rational basis” test, see, e.g., Williamson v. Lee Optical, 348 U.S. 483, 487-88, 75 S.Ct. 461, 464, 99 L.Ed. 563 (1955), but rather a more robust notion of “reasonableness” such as that applied in the Fourth Amendment context, see U.S. Const, amend. IV (prohibiting “unreasonable searches and seizures”).

In Burnside v. Byars, we articulated the way to determine whether a public school regulation that curtailed expression was reasonable:

[S]chool officials cannot ignore expressions of feelings with which they do not wish to contend. They cannot infringe on their students’ right to free and unrestricted expression as guaranteed to them under the First Amendment to the Constitution, where the exercise of such rights in the school buildings and schoolrooms do[es] not materially and substantially interfere with the requirements of appropriate discipline in the operation of the school.

363 F.2d at 749. Under the Burnside standard, student expression may unquestionably be regulated when doing so “contributes to the maintenance of order and decorum within the educational system.” Id. at 748; accord Tinker, 393 U.S. at 514, 89 .S.Ct. at 740 (holding that a school may not prohibit expressive activity unless there are “facts which might reasonably have led school authorities to forecast substantial disruption of or material interference with school activities”); Shanley, 462 F.2d at 969 (“The test for curtailing in-school exercise of expression is whether or not the expression or its method of exercise ‘materially and substantially’ interferes with the activities or discipline of the school.”). This doctrine allows school authorities to prohibit, among other things, “lewd, indecent, or offensive speech.... The First Amendment does not prevent the school officials from determining that to permit a vulgar and lewd speech such as respondent’s would undermine the school’s basic educational mission.” Bethel Sch. Dist. v. Fraser, 478 U.S. 675, 683, 685, 106 S.Ct. 3159, 3164-65, 92 L.Ed.2d 549 (1986); see also Healy v. James, 408 U.S. 169, 189, 92 S.Ct. 2338, 2350, 33 L.Ed.2d 266 (1972) (“[First Amendment] [a]ssociational activities need not be tolerated where they infringe reasonable campus rules, interrupt classes, or substantially interfere with the opportunity of other students to obtain an education.”).

However, in assessing the reasonableness. of regulations that tread upon expression, we cannot simply defer to the specter of disruption or the mere theoretical possibility of discord, or even some de minimis, insubstantial impact on classroom decorum. Particularly given the fact that young people are required by law to spend a-substantial portion of their lives in classrooms, student expression may not be suppressed simply because it -gives rise to some slight, easily overlooked disruption, including but not limited to “a showing of mild curiosity” by other students, see Burnside, 363 F.2d at 748, “discussion and comment” among students, Reineke v. Cobb Cty. Sch. Dist., 484 F.Supp. 1252, *12721261 (N.D.Ga.1980), or even some “hostile remarks” or “discussion outside of the classrooms” by other students, Tinker, 393 U.S. at 508, 514, 89 S.Ct. at 737, 740. For example, one district court correctly found that a teacher was unjustified in censoring an article in the school newspaper because it was “inconceivable that the use of the word ‘damn’ one time in the newspaper would have caused material and substantial interference with school activities.” Reineke, 484 F.Supp. at 1258.

The dissent concludes that Holloman’s gesture was unprotected because “[t]he students’ comments [to Allred after class] demonstrate that they at least focused their attention during a portion of the recitation of the Pledge on Holloman’s fist ... rather than on the planned curriculum of saying the Pledge.” This approach appears to ignore the principle discussed above that student expression must cause (or be likely to cause) a “material[] and substantial^” disruption, Burnside, 363 F.2d at 749, and more than a brief, easily overlooked, de minimis impact, before it may be curtailed. The dissent argues that Holloman’s act was “meant to compete for students’ attention.” The same can be said of any of the forms of student expression that have been found to be protected, including the wearing of armbands or buttons in class. A student expressing himself in those ways clearly intends to attract the other students’ attention and have them consider, however briefly, the meaning behind the symbolism. Indeed, if a student’s attention is never focused, if even for a moment, on the expression, it becomes pointless. Under the dissent’s approach, where schools may prohibit any speech or acts that do anything to distract a student’s mind — however briefly or insubstantially — from the planned curriculum, constitutional protection for student expression by definition would be eliminated.

This point was made quite clearly in Parducci v. Rutland, 316 F.Supp. 352 (M.D.Ala.1970), where a teacher was terminated because she assigned a short story that school administrators found offensive. The district court recognized that First Amendment freedoms in public schools, including a teacher’s right to academic freedom, could be constitutionally abridged under Tinker and Burnside only if there was a realistic threat that the conduct at issue would “materially and substantially interfere with the requirements of appropriate discipline in the operation of the school.” Id. at 355. The court, correctly applying our precedents, held:

Rather than there being a threatened or actual substantial disruption to the educational processes of the school, the evidence reflects that the assigning of the story was greeted with apathy by most of the students. Only three of plaintiffs students asked to be excused from the assignment. On this question of whether there was a material and substantial threat of disruption, the Principal testified at the School Board hearing that there was no indication that any of plaintiffs other 87 students were planning to disrupt the normal routine of the school. This Court now specifically finds and concludes that the conduct for which plaintiff was dismissed was not such that “would materially and substantially interfere with” reasonable requirements of discipline in the school.

Id. at 356. There is no evidence in the record to suggest that Holloman’s gesture caused any more disturbance or unrest than Parducci’s assignment. By focusing entirely on whether students may have been momentarily “distracted,” rather than on whether the distraction or disrup*1273tion was “material” or “substantial,” the dissent gives insufficient protection to students’ First Amendment rights and incorrectly applies our precedents.

While certain types of expression unquestionably cause enough of a threat of disruption to warrant suppression even before negative consequences occur, “undifferentiated fear or apprehension of disturbance is not enough to overcome the right to freedom of expression,” even in schools. Tinker, 393 U.S. at 508, 89 S.Ct. at 737. “[TJhere must be demonstrable factors that would give rise to any reasonable forecast by the school 'administration ■ of ‘substantial and material’ disruption of school activities before expression may be constitutionally restrained.” Shanley, 462 F.2d at 974; accord Center for Participant Ed. v. Marshall, 337 F.Supp. 126, 135 (N.D.Fla.1972) (suggesting that a “speculative fear” is insufficient to justify restrictions on student expression, but a “real and immediate fear of conduct potentially disruptive of the university routine” is enough).

We recognize that this test is more restrictive than the parsimonious interpretation of students’ First Amendment freedoms offered in Ferrell v. Dallas Indep. Sch. Dist., 392 F.2d 697 (5th Cir.1968). In Ferrell, a high school principal suspended students in a rock-and-roll band for growing their hair too long. This court upheld the principal’s regulation because the principal felt “that the length and style of the boys’ hair would cause commotion, trouble, distraction, and a disturbance in the school.... ” Id. at 699. To the degree this ruling is based on the principal’s abstract concerns, it is patently inconsistent with the principle articulated in Tinker, Burnside, and numerous other cases discussed throughout this Subsection, that there must be a real or substantial threat of actual disorder, as opposed to the mere possibility of one. We are bound to follow Burnside rather than Ferrell because it is the earlier case. See Local Union 48 Sheet Metal Workers v. S.L. Pappas & Co., 106 F.3d 970, 975 (11th Cir.1997). Similarly, we must follow Tinker rather than Ferrell because it is an intervening, inconsistent Supreme Court decision. See Lufkin v. McCallum, 956 F.2d 1104, 1107 (11th Cir.1992). Consequently, we apply the Tinker-Burnside doctrine in this case.

This Tinker-Burnside standard we reaffirm today was applied in Banks v. Bd. of Public Instr., 314 F.Supp. 285 (S.D.Fla.1970), vacated by 401 U.S. 988, 91 S.Ct. 1223, 28 L.Ed.2d 526 (1971), reinstated without published opinion by dist. ct. and aff'd, 450 F.2d 1103 (5th Cir.1971), a case similar to this one, where a student was suspended for failing to stand during the Pledge of Allegiance. The district court held, “The conduct of Andrew Banks in refusing to stand during the pledge ceremony constituted an expression of his religious beliefs and political opinions. His refusal to stand was no less a form of expression than the wearing of the black armband was to Mary Beth Tinker. He was exercising a right ‘akin to pure speech.’ ” Id. at 295. While at first glance it might seem like Banks’s right to be free from compelled speech was at issue in that case, the above quote from the district court makes clear that its ruling was not based on Banks’s First Amendment right to remain silent, but his First Amendment right to affirmatively express himself. Both the school administration and the court in that case perceived Banks’s act of remaining seated while everyone else was standing as an expressive act, rather than a mere refusal to speak. In upholding his right to act in such a way, the district court observed, “The unrefuted testimony clearly reflects that the plaintiffs refusal to stand has not caused any disruption in the educational process.” *1274Id.; cf. Jenkins v. Louisiana State Bd. of Educ., 506 F.2d 992 (5th Cir.1975) (upholding restrictions on college students’ activities because “[t]he actions of appellants resulted in a material disruption of the campus and of the rights of others. They were not protected by the First Amendment.”). We therefore find Banks to be fully consistent with the approach mandated by Burnside and Tinker, and conclude that Holloman’s gesture was sufficiently akin to Banks’s refusal to stand (in that neither had any real impact on class discipline) as to be entitled to First Amendment protection.

Our cases involving “freedom buttons” are perhaps even more instructive. In Blackwell v. Issaquena Cty. Bd. of Educ., 363 F.2d 749 (5th Cir.1966), black elementary school students wore “freedom buttons” to class in support of the civil rights movement.

[S]ome of these students were creating a disturbance by noisily talking in the hall when they were scheduled to be in class.... [Some students] accosted other students by pinning the buttons on them even though they did not ask for one. One of the students tried to put a button on a younger child who began crying. This activity created a state of confusion, disrupted class instruction, and resulted in a general breakdown of orderly discipline.

Id. at 750-52 (footnote omitted). We held that the principal did not violate the students’ constitutional rights by punishing them for their behavior and, under those circumstances, banning the buttons from the school. Such a restriction on student expression was justified, notwithstanding the First Amendment, because the “students conducted themselves in a disorderly manner, disrupted classroom procedure, interfered with the proper decorum and discipline of the school and disturbed other students who did not wish to participate in the wearing of the buttons.” Id. at 753.