*843Justice O’Connor, Justice Kennedy, and Justice Sou-ter

announced the judgment of the Court and delivered the opinion of the Court with respect to Parts I, II, III, V-A, *844V-C, and VI, an opinion with respect to Part V-E,

in which Justice Stevens joins, and an opinion with respect to Parts IV, V-B, and V-D.

I

Liberty finds no refuge in a jurisprudence of doubt. Yet 19 years after our holding that the Constitution protects a woman’s right to terminate her pregnancy in its early stages, Roe v. Wade, 410 U. S. 113 (1973), that definition of liberty is still questioned. Joining the respondents as amicus curiae, the United States, as it has done in five other cases in the last decade, again asks us to overrule Roe. See Brief for Respondents 104-117; Brief for United States as Amicus Curiae 8.

At issue in these cases are five provisions of the Pennsylvania Abortion Control Act of 1982, as amended in 1988 and 1989. 18 Pa. Cons. Stat. §§3203-3220 (1990). Relevant portions of the Act are set forth in the Appendix. Infra, at 902. The Act requires that a woman seeking an abortion give her informed consent prior to the abortion procedure, and specifies that she be provided with certain information at least 24 hours before the abortion is performed. § 3205. For a minor to obtain an abortion, the Act requires the informed consent of one of her parents, but provides for a judicial bypass option if the minor does not wish to or cannot obtain a parent’s consent. § 3206. Another provision of the Act requires that, unless certain exceptions apply, a married woman seeking an abortion must sign a statement indicating that she has notified her husband of her intended abortion. §3209. The Act exempts compliance with these three requirements in the event of a “medical emergency,” which is defined in §3203 of the Act. See §§3203, 3205(a), 3206(a), 3209(c). In addition to the above provisions regulating the performance of abortions, the Act imposes certain reporting requirements on facilities that provide abortion services. §§ 3207(b), 3214(a), 3214(f).

*845Before any of these provisions took effect, the petitioners, who are five abortion clinics and one physician representing himself as well as a class of physicians who provide abortion services, brought this suit seeking declaratory and injunctive relief. Each provision was challenged as unconstitutional on its face. The District Court entered a preliminary injunction against the enforcement of the regulations, and, after a 3-day bench trial, held all the provisions at issue here unconstitutional, entering a permanent injunction against Pennsylvania’s enforcement of them. 744 F. Supp. 1323 (ED Pa. 1990). The Court of Appeals for the Third Circuit affirmed in part and reversed in part, upholding all of the regulations except for the husband notification requirement. 947 F. 2d 682 (1991). We granted certiorari. 502 U. S. 1056 (1992).

The Court of Appeals found it necessary to follow an elaborate course of reasoning even to identify the first premise to use to determine whether the statute enacted by Pennsylvania meets constitutional standards. See 947 F. 2d, at 687-698. And at oral argument in this Court, the attorney for the parties challenging the statute took the position that none of the enactments can be upheld without overruling Roe v. Wade. Tr. of Oral Arg. 5-6. We disagree with that analysis; but we acknowledge that our decisions after Roe cast doubt upon the meaning and reach of its holding. Further, The Chief Justice admits that he would overrule the central holding of Roe and adopt the rational relationship test as the sole criterion of constitutionality. See post, at 944, 966. State and federal courts as well as legislatures throughout the Union must have guidance as they seek to address this subject in conformance with the Constitution. Given these premises, we find it imperative to review once more the principles that define the rights of the woman and the legitimate authority of the State respecting the termination of pregnancies by abortion procedures.

After considering the fundamental constitutional questions resolved by Roe, principles of institutional integrity, *846and the rule of stare decisis, we are led to conclude this: the essential holding of Roe v. Wade should be retained and once again reaffirmed.

It must be stated at the outset and with clarity that Roe’s essential holding, the holding we reaffirm, has three parts. First is a recognition of the right of the woman to choose to have an abortion before viability and to obtain it without undue interference from the State. Before viability, the State’s interests are not strong enough to support a prohibition of abortion or the imposition of a substantial obstacle to the woman’s effective right to elect the procedure. Second is a confirmation of the State’s power to restrict abortions after fetal viability, if the law contains exceptions for pregnancies which endanger the woman’s life or health. And third is the principle that the State has legitimate interests from the outset of the pregnancy in protecting the health of the woman and the life of the fetus that may become a child. These principles do not contradict one another; and we adhere to each.

II

Constitutional protection of the woman’s decision to terminate her pregnancy derives from the Due Process Clause of the Fourteenth Amendment. It declares that no State shall “deprive any person of life, liberty, or property, without due process of law.” The controlling word in the cases before us is “liberty.” Although a literal reading of the Clause might suggest that it governs only the procedures by which a State may deprive persons of liberty, for at least 105 years, since Mugler v. Kansas, 123 U. S. 623, 660-661 (1887), the Clause has been understood to contain a substantive component as well, one “barring certain government actions regardless of the fairness of the procedures used to implement them.” Daniels v. Williams, 474 U. S. 327, 331 (1986). As Justice Brandéis (joined by Justice Holmes) observed, “[djespite arguments to the contrary which had seemed to me persuasive, it is settled that the due process clause of the Fourteenth *847Amendment applies to matters of substantive law as well as to matters of procedure. Thus all fundamental rights comprised within the term liberty are protected by the Federal Constitution from invasion by the States.” Whitney v. California, 274 U. S. 357, 373 (1927) (concurring opinion). “[TJhe guaranties of due process, though having their roots in Magna Carta’s ‘per legem ter rae’ and considered as procedural safeguards ‘against executive usurpation and tyranny/ have in this country ‘become bulwarks also against arbitrary legislation.’ ” Poe v. Ullman, 367 U. S. 497, 541 (1961) (Harlan, J., dissenting from dismissal on jurisdictional grounds) (quoting Hurtado v. California, 110 U. S. 516, 532 (1884)).

The most familiar of the substantive liberties protected by the Fourteenth Amendment are those recognized by the Bill of Rights. We have held that the Due Process Clause of the Fourteenth Amendment incorporates most of the Bill of Rights against the States. See, e. g., Duncan v. Louisiana, 391 U. S. 145, 147-148 (1968). It is tempting, as a means of curbing the discretion of federal judges, to suppose that liberty encompasses no more than those rights already guaranteed to the individual against federal interference by the express provisions of the first eight Amendments to the Constitution. See Adamson v. California, 332 U. S. 46, 68-92 (1947) (Black, J., dissenting). But of course this Court has never accepted that view.

It is also tempting, for the same reason, to suppose that the Due Process Clause protects only those practices, defined at the most specific level, that were protected against government interference by other rules of law when the Fourteenth Amendment was ratified. See Michael H. v. Gerald D., 491 U. S. 110, 127-128, n. 6 (1989) (opinion of Scalia, J.). But such a view would be inconsistent with our law. It is a promise of the Constitution that there is a realm of personal liberty which the government may not enter. We have vindicated this principle before. Marriage is mentioned nowhere in the Bill of Rights and interracial marriage was file-*848gal in most States in the 19th century, but the Court was no doubt correct in finding it to be an aspect of liberty protected against state interference by the substantive component of the Due Process Clause in Loving v. Virginia, 388 U. S. 1, 12 (1967) (relying, in an opinion for eight Justices, on the Due Process Clause). Similar examples may be found in Turner v. Safley, 482 U. S. 78, 94-99 (1987); in Carey v. Population Services International, 431 U. S. 678, 684-686 (1977); in Griswold v. Connecticut, 381 U. S. 479, 481-482 (1965), as well as in the separate opinions of a majority of the Members of the Court in that case, id., at 486-488 (Goldberg, J., joined by Warren, C. J., and Brennan, J., concurring) (expressly relying on due process), id., at 500-502 (Harlan, J., concurring in judgment) (same), id., at 502-507 (White, J., concurring in judgment) (same); in Pierce v. Society of Sisters, 268 U. S. 510, 534-535 (1925); and in Meyer v. Nebraska, 262 U. S. 390, 399-403 (1923).

Neither the Bill of Rights nor the specific practices of States at the time of the adoption of the Fourteenth Amendment marks the outer limits of the substantive sphere of liberty which the Fourteenth Amendment protects. See U. S. Const., Arndt. 9. As the second Justice Harlan recognized:

“[T]he full scope of the liberty guaranteed by the Due Process Clause cannot be found in or limited by the precise terms of the specific guarantees elsewhere provided in the Constitution. This ‘liberty’ is not a series of isolated points pricked out in terms of the taking of property; the freedom of speech, press, and religion; the right to keep and bear arms; the freedom from unreasonable searches and seizures; and so on. It is a rational continuum which, broadly speaking, includes a freedom from all substantial arbitrary impositions and purposeless restraints, . . . and which also recognizes, what a reasonable and sensitive judgment must, that certain interests require particularly careful scrutiny of the state needs asserted to justify their abridgment.” Poe v. *849UUman, supra, at 543 (opinion dissenting from dismissal on jurisdictional grounds).

Justice Harlan wrote these words in addressing an issue the full Court did not reach in Poe v. Ullman, but the Court adopted his position four Terms later in Griswold v. Connecticut, supra. In Griswold, we held that the Constitution does not permit a State to forbid a married couple to use contraceptives. That same freedom was later guaranteed, under the Equal Protection Clause, for unmarried couples. See Eisenstadt v. Baird, 405 U. S. 438 (1972). Constitutional protection was extended to the sale and distribution of contraceptives in Carey v. Population Services International, supra. It is settled now, as it was when the Court heard arguments in Roe v. Wade, that the Constitution places limits on a State’s right to interfere with a person’s most basic decisions about family and parenthood, see Carey v. Population Services International, supra; Moore v. East Cleveland, 431 U. S. 494 (1977); Eisenstadt v. Baird, supra; Loving v. Virginia, supra; Griswold v. Connecticut, supra; Skinner v. Oklahoma ex rel. Williamson, 316 U. S. 535 (1942); Pierce v. Society of Sisters, supra; Meyer v. Nebraska, supra, as well as bodily integrity, see, e. g., Washington v. Harper, 494 U. S. 210, 221-222 (1990); Winston v. Lee, 470 U. S. 753 (1985); Rochin v. California, 342 U. S. 165 (1952).

The inescapable fact is that adjudication of substantive due process claims may call upon the Court in interpreting the Constitution to exercise that same capacity which by tradition courts always have exercised: reasoned judgment. Its boundaries are not susceptible of expression as a simple rule. That does not mean we are free to invalidate state policy choices with which we disagree; yet neither does it permit us to shrink from the duties of our office. As Justice Harlan observed:

“Due process has not been reduced to any formula; its content cannot be determined by reference to any code. *850The best that can be said is that through the course of this Court’s decisions it has represented the balance which our Nation, built upon postulates of respect for the liberty of the individual, has struck between that liberty and the demands of organized society. If the supplying of content to this Constitutional concept has of necessity been a rational process, it certainly has not been one where judges have felt free to roam where unguided speculation might take them. The balance of which I speak is the balance struck by this country, having regard to what history teaches are the traditions from which it developed as well as the traditions from which it broke. That tradition is a living thing. A decision of this Court which radically departs from it eould not long survive, while a decision which builds on what has survived is likely to be sound. No formula could serve as a substitute, in this area, for judgment and restraint.” Poe v. Ullman, 367 U. S., at 542 (opinion dissenting from dismissal on jurisdictional grounds).

See also Rochin v. California, supra, at 171-172 (Frankfurter, J., writing for the Court) (“To believe that this judicial exercise of judgment could be avoided by freezing ‘due process of law’ at some fixed stage of time or thought is to suggest that the most important aspect of constitutional adjudication is a function for inanimate machines and not for judges”).

Men and women of good conscience can disagree, and we suppose some always shall disagree, about the profound moral and spiritual implications of terminating a pregnancy, even in its earliest stage. Some of us as individuals find abortion offensive to our most basic principles of morality, but that cannot control our decision. Our obligation is to define the liberty of all, not to mandate our own moral code. The underlying constitutional issue is whether the State can resolve these philosophic questions in such a definitive way that a woman lacks all choice in the matter, except perhaps *851in those rare circumstances in which the pregnancy is itself a danger to her own life or health, or is the result of rape or incest.

It is conventional constitutional doctrine that where reasonable people disagree the government can adopt one position or the other. See, e. g., Ferguson v. Skrupa, 372 U. S. 726 (1963); Williamson v. Lee Optical of Okla., Inc., 348 U. S. 483 (1955). That theorem, however, assumes a state of affairs in which the choice does not intrude upon a protected liberty. Thus, while some people might disagree about whether or not the flag should be saluted, or disagree about the proposition that it may not be defiled, we have ruled that a State may not compel or enforce one view or the other. See West Virginia Bd. of Ed. v. Barnette, 319 U. S. 624 (1943); Texas v. Johnson, 491 U. S. 397 (1989).

Our law affords constitutional protection to personal decisions relating to marriage, procreation, contraception, family relationships, child rearing, and education. Carey v. Population Services International, 431 U. S., at 685. Our cases recognize “the right of the individual, married or single, to be free from unwarranted governmental intrusion into matters so fundamentally affecting a person as the decision whether to bear or beget a child.” Eisenstadt v. Baird, supra, at 453 (emphasis in original). Our precedents “have respected the private realm of family life which the state cannot enter.” Prince v. Massachusetts, 321 U. S. 158, 166 (1944). These matters, involving the most intimate and personal choices a person may make in a lifetime, choices central to personal dignity and autonomy, are central to the liberty protected by the Fourteenth Amendment. At the heart of liberty is the right to define one’s own concept of existence, of meaning, of the universe, and of the mystery of human life. Beliefs about these matters could not define the attributes of personhood were they formed under compulsion of the State.

*852These considerations begin our analysis of the woman’s interest in terminating her pregnancy but cannot end it, for this reason: though the abortion decision may originate within the zone of conscience and belief, it is more than a philosophic exercise. Abortion is a unique act. It is an act fraught with consequences for others: for the woman who must live with the implications of her decision; for the persons who perform and assist in the procedure; for the spouse, family, and society which must confront the knowledge that these procedures exist, procedures some deem nothing short of an act of violence against innocent human life; and, depending on one’s beliefs, for the life or potential life that is aborted. Though abortion is conduct, it does not follow that the State is entitled to proscribe it in all instances. That is because the liberty of the woman is at stake in a sense unique to the human condition and so unique to the law. The mother who carries a child to full term is subject to anxieties, to physical constraints, to pain that only she must bear. That these sacrifices have from the beginning of the human race been endured by woman with a pride that ennobles her in the eyes of others and gives to the infant a bond of love cannot alone be grounds for the State to insist she make the sacrifice. Her suffering is too intimate and personal for the State to insist, without more, upon its own vision of the woman’s role, however dominant that vision -has been in the course of our history and our culture. The destiny of the woman must be shaped to a large extent on her own conception of her spiritual imperatives and her place in society.

It should be recognized, moreover, that in some critical respects the abortion decision is of the same character as the decision to use contraception, to which Griswold v. Connecticut, Eisenstadt v. Baird, and Carey v. Population Services International afford constitutional protection. We have no doubt as to the correctness of those decisions. They support *853the reasoning in Roe relating to the woman’s liberty because they involve personal decisions concerning not only the meaning of procreation but also human responsibility and respect for it. As with abortion, reasonable people will have differences of opinion about these matters. One view is based on such reverence for the wonder of creation that any pregnancy ought to be welcomed and carried to full term no matter how difficult it will be to provide for the child and ensure its well-being. Another is that the inability to provide for the nurture and care of the infant is a cruelty to the child and an anguish to thé parent. These are intimate views with infinite variations, and their deep, personal character underlay our decisions in Griswold, Eisenstadt, and Carey. The same concerns are present when the woman confronts the reality that, perhaps despite her attempts to avoid it, she has become pregnant.

It was this dimension of personal liberty that Roe sought to protect, and its holding invoked the reasoning and the tradition of the precedents we have discussed, granting protection to substantive liberties of the person. Roe was, of course, an extension of those cases and, as the decision itself indicated, the separate States could act in some degree to further their own legitimate interests in protecting prenatal life. The extent to which the legislatures of the States might aet to outweigh the interests of the woman in choosing to terminate her pregnancy was a subject of debate both in Roe itself and in decisions following it.

While we appreciate the weight of the arguments made on behalf of the State in the cases before us, arguments which in their ultimate formulation conclude that Roe should be overruled, the reservations any of us may have in reaffirming the central holding of Roe are outweighed by the explication of individual liberty we have given combined with the force of stare decisis. We turn now to that doctrine.

*854hH I — I

A

The obligation to follow precedent begins with necessity, and a contrary necessity marks its outer limit. With Cardozo, we recognize that no judicial system could do society's work if it eyed each issue afresh in every case that raised it. See B. Cardozo, The Nature of the Judicial Process 149 (1921). Indeed, the very concept of the rule of law underlying our own Constitution requires such continuity over time that a respect for precedent is, by definition, indispensable. See Powell, Stare Decisis and Judicial Restraint, 1991 Journal of Supreme Court History 13,16. At the other extreme, a different necessity would make itself felt if a prior judicial ruling should come to be seen so clearly as error that its enforcement was for that very reason doomed.

Even when the decision to overrule a prior case is not, as in the rare, latter instance, virtually foreordained, it is common wisdom that the rule of stare decisis is not an “inexorable command,” and certainly it is not such in every constitutional case, see Burnet v. Coronado Oil & Gas Co., 285 U. S. 393, 405-411 (1932) (Brandéis, J., dissenting). See also Payne v. Tennessee, 501 U. S. 808, 842 (1991) (Souter, J., joined by Kennedy, J., concurring); Arizona v. Rumsey, 467 U. S. 203, 212 (1984). Rather, when this Court reexamines a prior holding, its judgment is customarily informed by a series of prudential and pragmatic considerations designed to test the consistency of overruling a prior decision with the ideal of the rule of law, and to gauge the respective costs of reaffirming and overruling a prior case. Thus, for example, we may ask whether the rule has proven to be intolerable simply in defying practical workability, Swift & Co. v. Wickham, 382 U. S. 111, 116 (1965); whether the rule is subject to a kind of reliance that would lend a special hardship to the consequences of overruling and add inequity to the cost of repudiation, e. g., United States v. Title Ins. & Trust *855Co., 265 U. S. 472, 486 (1924); whether related principles of law have so far developed as to have left the old rule no more than a remnant of abandoned doctrine, see Patterson v. McLean Credit Union, 491 U. S. 164, 173-174 (1989); or whether facts have so changed, or come to be seen so differently, as to have robbed the old rule of significant application or justification, e.g., Burnet, supra, at 412 (Brandéis, J., dissenting).

So in this case we may enquire whether Roe’s central rule has been found unworkable; whether the rule’s limitation on state power could be removed without serious inequity to those who have relied upon it or significant damage to the stability of the society governed by it; whether the law’s growth in the intervening years has left Roe’s central rule a doctrinal anachronism discounted by society; and whether Roe’s premises of fact have so far changed in the ensuing two decades as to render its central holding somehow irrelevant or unjustifiable in dealing with the issue it addressed.

1

Although Roe has engendered opposition, it has in no sense proven “unworkable,” see Garcia v. San Antonio Metropolitan Transit Authority, 469 U. S. 528, 546 (1985), representing as it does a simple limitation beyond which a state law is unenforceable. While Roe has, of course, required judicial assessment of state laws affecting the exercise of the choice guaranteed against government infringement, and although the need for such review will remain as a consequence of today’s decision, the required determinations fall within judicial competence.

2

The inquiry into reliance counts the cost of a rule’s repudiation as it would fall on those who have relied reasonably on the rule’s continued application. Since the classic case for weighing reliance heavily in favor of following the earlier rule occurs in the commercial context, see Payne v. Tennes*856see, supra,, at 828, where advance planning of great precision is most obviously a necessity, it is no cause for surprise that some would find no reliance worthy of consideration in support of Roe.

While neither respondents nor their amici in so many words deny that the abortion right invites some reliance prior to its actual exercise, one can readily imagine an argument stressing the dissimilarity of this case to one involving property or contract. Abortion is customarily chosen as an unplanned response to the consequence of unplanned activity or to the failure of conventional birth control, and except on the assumption that no intercourse would have occurred but for Roe’s holding, such behavior may appear to justify no reliance claim. Even if reliance could be claimed on that unrealistic assumption, the argument might run, any reliance interest would be de minimis. This argument would be premised on the hypothesis that reproductive planning could take virtually immediate account of any sudden restoration of state authority to ban abortions.

To eliminate the issue of relianee that easily, however, one would need to limit cognizable reliance to specific instances of sexual activity. But to do this would be simply to refuse to face the fact that for two decades of economic and social developments, people have organized intimate relationships and made choices that define their views of themselves and their places in society, in relianee on the availability of abortion in the event that contraception should fail. The ability of women to participate equally in the economic and social life of the Nation has been facilitated by their ability to control their reproductive lives. See, e. g., R. Petehesky, Abortion and Woman’s Choice 109, 138, n. 7 (rev. ed. 1990). The Constitution serves human values, and while the effect of reliance on Roe cannot be exactly measured, neither can the certain cost of overruling Roe for people who have ordered their thinking and living around that case be dismissed.

*8573

No evolution of legal principle has left Roe’s doctrinal footings weaker than they were in 1973. No development of constitutional law since the ease was decided has implicitly or explicitly left Roe behind as a mere survivor of obsolete constitutional thinking.

It will be recognized, of course, that Roe stands at an intersection of two lines of decisions, but in whichever doctrinal category one reads the case, the result for present purposes will be the same. The Roe Court itself placed its holding in the succession of cases most prominently exemplified by Griswold v. Connecticut, 381 U. S. 479 (1965). See Roe, 410 U. S., at 152-153. When it is so seen, Roe is clearly in no jeopardy, since subsequent constitutional developments have neither disturbed, nor do they threaten to diminish, the scope of recognized protection accorded to the liberty relating to intimate relationships, the family, and decisions about whether or not to beget or bear a child. See, e. g., Carey v. Population Services International, 431 U. S. 678 (1977); Moore v. East Cleveland, 431 U. S. 494 (1977).

Roe, however, may be seen not only as an exemplar of Griswold liberty but as a rule (whether or not mistaken) of personal autonomy and bodily integrity, with doctrinal affinity to cases recognizing limits on governmental power to mandate medical treatment or to bar its rejection. If so, our cases since Roe accord with Roe’s view that a State’s interest in the protection of life falls short of justifying any plenary override of individual liberty claims. Cruzan v. Director, Mo. Dept. of Health, 497 U. S. 261, 278 (1990); cf., e. g., Riggins v. Nevada, 504 U. S. 127, 135 (1992); Washington v. Harper, 494 U. S. 210 (1990); see also, e. g., Rochin v. California, 342 U. S. 165 (1952); Jacobson v. Massachusetts, 197 U. S. 11, 24-30 (1905).

Finally, one could classify Roe as sui generis. If the case is so viewed, then there clearly has been no erosion of its central determination. The original holding resting on the *858concurrence of seven Members of the Court in 1973 was expressly affirmed by a majority of six in 1983, see Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416 (Akron I), and by a majority of five in 1986, see Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S. 747, expressing adherence to the constitutional ruling despite legislative efforts in some States to test its limits. More recently, in Webster v. Reproductive Health Services, 492 U. S. 490 (1989), although two of the present authors questioned the trimester framework in a way consistent with our judgment today, see id., at 518 (Rehnquist, C. J., joined by White and Kennedy, JJ.); id., at 529 (O’Connor, J., concurring in part and concurring in judgment), a majority of the Court either decided to reaffirm or declined to address the constitutional validity of the central holding of Roe. See Webster, 492 U. S., at 521 (Rehnquist, C. J., joined by White and Kennedy, JJ.); id., at 525-526 (O’Connor, J., concurring in part and concurring in judgment); id., at 537, 553 (Blackmun, J., joined by Brennan and Marshall, JJ., concurring in part and dissenting in part); id., at 561-563 (Stevens, J., concurring in part and dissenting in part).

Nor will courts building upon Roe be likely to hand down erroneous decisions as a consequence. Even on the assumption that the central holding of Roe was in error, that error would go only to the strength of the state interest in fetal protection, not to the recognition afforded by the Constitution to the woman’s liberty. The latter aspect of the decision fits comfortably within the framework of the Court’s prior decisions, including Skinner v. Oklahoma ex rel. Williamson, 316 U. S. 535 (1942); Griswold, supra; Loving v. Virginia, 388 U. S. 1 (1967); and Eisenstadt v. Baird, 405 U. S. 438 (1972), the holdings of which are “not a series of isolated points,” but mark a “rational continuum.” Poe v. Ullman, 367 U. S., at 543 (Harlan, J, dissenting). As we described in *859Carey v. Population Services International, supra, the liberty which encompasses those decisions

“includes ‘the interest in independence in making certain kinds of important decisions/ While the outer limits of this aspect of [protected liberty] have not been marked by the Court, it is clear that among the decisions that an individual may make without unjustified government interference are personal decisions ‘relating to marriage, procreation, contraception, family relationships, and child rearing and education/ ” 431 U. S., at 684-685 (citations omitted).

The soundness of this prong of the Roe analysis is apparent from a consideration of the alternative. If indeed the woman’s interest in deciding whether to bear and beget a child had not been recognized as in Roe, the State might as readily restrict a woman’s right to choose to carry a pregnancy to term as to terminate it, to further asserted state interests in population control, or eugenics, for example. Yet Roe has been sensibly relied upon to counter any such suggestions. E. g., Arnold v. Board of Education of Escambia County, Ala., 880 F. 2d 305, 311 (CA11 1989) (relying upon Roe and concluding that government officials violate the Constitution by coercing a minor to have an abortion); Avery v. County of Burke, 660 F. 2d 111, 115 (CA4 1981) (county agency inducing teenage girl to undergo unwanted sterilization on the basis of misrepresentation that she had sickle cell trait); see also In re Quinlan, 70 N. J. 10, 355 A. 2d 647 (relying on Roe in finding a right to terminate medical treatment), cert, denied sub nom. Garger v. New Jersey, 429 U. S. 922 (1976)). In any event, because Roe’s scope is confined by the fact of its concern with postconeeption potential life, a concern otherwise likely to be implicated only by some forms of contraception protected independently under Griswold and later cases, any error in Roe is unlikely to have serious ramifications in fixture cases.

*8604

We have seen how time has overtaken some of Roe’s factual assumptions: advances in maternal health care allow for abortions safe to the mother later in pregnancy than was true in 1973, see Akron I, supra, at 429, n. 11, and advances in neonatal care have advanced viability to a point somewhat earlier. Compare Roe, 410 U. S., at 160, with Webster, supra, at 515-616 (opinion of Rehnquist, C. J.); see Akron I, 462 U. S., at 457, and n. 5 (O’Connor, J., dissenting). But these facts go only to the scheme of time limits on the realization of competing interests, and the divergences from the factual premises of 1973 have no bearing on the validity of Roe’s central holding, that viability marks the earliest point at which the State’s interest in fetal life is constitutionally adequate to justify a legislative ban on nontherapeutic abortions. The soundness or unsoundness of that constitutional judgment in no sense turns on whether viability occurs at approximately 28 weeks, as was usual at the time of Roe, at 23 to 24 weeks, as it sometimes does today, or at some moment even slightly earlier in pregnancy, as it may if fetal respiratory capacity can somehow be enhanced in the future. Whenever it may occur, the attainment of viability may continue to serve as the critical fact, just as it has done since Roe was decided; which is to say that no change in Roe’s factual underpinning has left its central holding obsolete, and none supports an argument for overruling it.

5

The sum of the precedential enquiry to this point Shows Roe’s underpinnings unweakened in any way affecting its central holding. While it has engendered disapproval, it has not been unworkable. An entire generation has come of age free to assume Roe’s concept of liberty in defining the capacity of women to act in society, and to make reproductive decisions; no erosion of principle going to liberty or personal autonomy has left Roe’s central holding a doctrinal remnant; *861Roe portends no developments at odds with other precedent for the analysis of personal liberty; and no changes of fact have rendered viability more or less appropriate as the point at which the balance of interests tips. Within the bounds of normal stare decisis analysis, then, and subject to the considerations on which it customarily turns, the stronger argument is for affirming Roe’s central holding, with whatever degree of personal reluctance any of us may have, not for overruling it.

B

In a less significant case, stare decisis analysis could, and would, stop at the point we have reached. But the sustained and widespread debate Roe has provoked calls for some comparison between that case and others of comparable dimension that have responded to national controversies and taken on the impress of the controversies addressed. Only two such decisional lines from the past century present themselves for examination, and in each instance the result reached by the Court accorded with the principles we apply today.

The first example is that line of cases identified with Lochner v. New York, 198 U. S. 45 (1905), which imposed substantive limitations on legislation limiting economic autonomy in favor of health and welfare regulation, adopting, in Justice Holmes’s view, the theory of laissez-faire. Id., at 75 (dissenting opinion). The Lochner decisions were exemplified by Adkins v. Children’s Hospital of District of Columbia, 261 U. S. 525 (1923), in which this Court held it to be an infringement of constitutionally protected liberty of contract to require the employers of adult women to satisfy minimum wage standards. Fourteen years later, West Coast Hotel Co. v. Parrish, 300 U. S. 379 (1937), signaled the demise of Lochner by overruling Adkins. In the meantime, the Depression had come and, with it, the lesson that seemed unmistakable to most people by 1937, that the interpretation of contractual freedom protected in Adkins rested on funda*862mentally false factual assumptions about the capacity of a relatively unregulated market to satisfy minimal levels of human welfare. See West Coast Hotel Co., supra, at 399. As Justice Jackson wrote of the constitutional crisis of 1937 shortly before he came on the bench: “The older world of laissez faire was recognized everywhere outside the Court to be dead.” The Struggle for Judicial Supremacy 85 (1941). The facts upon which the earlier case had premised a constitutional resolution of social controversy had proven to be untrue, and history’s demonstration of their untruth not only justified but required the new choice of constitutional principle that West Coast Hotel announced. Of course, it was true that the Court lost something by its misperception, or its lack of prescience, and the Court-packing crisis only magnified the loss; but the clear demonstration that the facts of economic life were different from those previously assumed warranted the repudiation of the old law.

The second comparison that 20th century history invites is with the cases employing the separate-but-equal rule for applying the Fourteenth Amendment’s equal protection guarantee. They began with Plessy v. Ferguson, 163 U. S. 537 (1896), holding that legislatively mandated racial segregation in public transportation works no denial of equal protection, rejecting the argument that racial separation enforced by the legal machinery of American society treats the black race as inferior. The Plessy Court considered “the underlying fallacy of the plaintiff’s argument to consist in the assumption that the enforced separation of the two races stamps the colored race with a badge of inferiority. If this be so, it is not by reason of anything found in the act, but solely because the colored race chooses to put that construction upon it.” Id., at 551. Whether, as a matter of historical fact, the Justices in the Plessy majority believed this or not, see id., at 557, 562 (Harlan, J., dissenting), this understanding of the implication of segregation was the stated justification for the Court’s opinion. But this understanding of *863the facts and the rule it was stated to justify were repudiated in Brown v. Board of Education, 347 U. S. 483 (1954) (Brown I). As one commentator observed, the question before the Court in Brown was “whether discrimination inheres in that segregation which is imposed by law in the twentieth century in certain specific states in the American Union. And that question has meaning and can find an answer only on the ground of history and of common knowledge about the facts of life in the times and places aforesaid.” Black, The Lawfulness of the Segregation Decisions, 69 Yale L. J. 421, 427 (1960).

The Court in Brown addressed these facts of life by observing that whatever may have been the understanding in Plessy’s time of the power of segregation to stigmatize those who were segregated with a “badge of inferiority,” it was clear by . 1954 that legally sanctioned segregation had just such an effect, to the point that racially separate public educational facilities were deemed inherently unequal. 347 U. S., at 494-495. Society’s understanding of the facts upon which a constitutional ruling was sought in 1954 was thus fundamentally different from the basis claimed for the decision in 1896. While we think Plessy was wrong the day it was decided, see Plessy, supra, at 552-564 (Harlan, J., dissenting), we must also recognize that the Plessy Court’s explanation for its decision was so clearly at odds with the facts apparent to the Court in 1954 that the decision to reexamine Plessy was on this ground alone not only justified but required.

West Coast Hotel and Brown each rested on facts, or an understanding of facts, changed from those which furnished the claimed justifications for the earlier constitutional resolutions. Each case was comprehensible as the Court’s response to facts that the country could understand, or had come to understand already, but which the Court of an earlier day, as its own declarations disclosed, had not been able to perceive. As the decisions were thus comprehensible *864they were also defensible, not merely as the victories of one doctrinal school over another by dint of numbers (victories though they were), but as applications of constitutional principle to facts as they had not been seen by the Court before. In constitutional adjudication as elsewhere in life, changed circumstances may impose new obligations, and the thoughtful part of the Nation could accept each decision to overrule a prior ease as a response to the Court’s constitutional duty.

Because the cases before us present no such occasion it could be seen as no such response. Because neither the factual underpinnings of Roe’s central holding nor our understanding of it has changed (and because no other indication of weakened precedent has been shown), the Court could not pretend to be reexamining the prior law with any justification beyond a present doctrinal disposition to come out differently from the Court of 1973. To overrule prior law for no other reason than that would run counter to the view repeated in our eases, that a decision to overrule should rest on some special reason over and above the belief that a prior ease was wrongly decided. See, e. g., Mitchell v. W. T. Grant Co., 416 U. S. 600, 636 (1974) (Stewart, J., dissenting) (“A basic change in the law upon a ground no firmer than a change in our membership invites the popular misconception that this institution is little different from the two political branches of the Government. No misconception could do more lasting injury to this Court and to the system of law which it is our abiding mission to serve”); Mapp v. Ohio, 367 U. S. 643, 677 (1961) (Harlan, J., dissenting).

C

The examination of the conditions justifying the repudiation of Adkins by West Coast Hotel and Plessy by Brown is enough to suggest the terrible price that would have been paid if the Court had not overruled as it did. In the present cases, however, as our analysis to this point makes clear, the terrible price would be paid for overruling. Our analysis *865would not be complete, however, without explaining why-overruling Roe’s central holding would not only reach an unjustifiable result under principles of stare decisis, but would seriously weaken the Court’s capacity to exercise the judicial power and to function as the Supreme Court of a Nation dedicated to the rule of law. To understand why this would be so it is necessary to understand the source of this Court’s authority, the conditions necessary for its preservation, and its relationship to the country’s understanding of itself as a constitutional Republic.

The root of American governmental power is revealed most clearly in the instance of the power conferred by the Constitution upon the-Judiciary of the United States and specifically upon this Court. As Americans of each succeeding generation are rightly told, the Court cannot buy support for its decisions by spending money and, except to a minor degree, it cannot independently coerce obedience to its decrees. The Court’s power lies, rather, in its legitimacy, a product of substance and perception that shows itself in the people’s acceptance of the Judiciary as fit to determine what the Nation’s law means and to declare what it demands.

The underlying substance of this legitimacy is of course the warrant for the Court’s decisions in the Constitution and the lesser sources of legal principle on which the Court draws. That substance is expressed in the Court’s opinions, and our contemporary understanding is such that a decision without principled justification would be no judicial act at all. But even when justification is furnished by apposite legal principle, something more is required. Because not every conscientious claim of principled justification will be accepted as such, the justification claimed must be beyond dispute. The Court must take care to speak and act in ways that allow people to accept its decisions on the terms the Court claims for them, as grounded truly in principle, not as compromises with social and political pressures having, as such, no bearing on the principled choices that the Court is *866obliged to make. Thus, the Court’s legitimacy depends on making legally principled decisions under circumstances in which their principled character is sufficiently plausible to be accepted by the Nation.

The need for principled action to be perceived as such is implicated to some degree whenever this, or any other appellate court, overrules a prior case. This is not to say, of course, that this Court cannot give a perfectly satisfactory explanation in most cases. People understand that some of the Constitution’s language is hard to fathom and that the Court’s Justices are sometimes able to perceive significant facts or to understand principles of law that* eluded their predecessors and that justify departures from existing decisions. However upsetting it may be to those most directly affected when one judicially derived rule replaces another, the country can accept some correction of error without necessarily questioning the legitimacy of the Court.

In two circumstances, however, the Court would almost certainly fail to receive the benefit of the doubt in overruling prior cases. There is, first, a point beyond which frequent overruling would overtax the country’s belief in the Court’s good faith. Despite the variety of reasons that may inform and justify a decision to overrule, we cannot forget that such a decision is usually perceived (and perceived correctly) as, at the least, a statement that a prior decision was wrong. There is a limit to the amount of error that can plausibly be imputed to prior Courts. If that limit should be exceeded, disturbance of prior rulings would be taken as evidence that justifiable reexamination of principle had given way to drives for particular results in the short term. The legitimacy of the Court would fade with the frequency of its vacillation.

That first circumstance can be described as hypothetical; the second is to the point here and now. Where, in the performance of its judicial duties, the Court decides a ease in such a way as to resolve the sort of intensely divisive controversy reflected in Roe and those rare, comparable cases, its *867decision has a dimension that the resolution of the normal case does not carry. It is the dimension present whenever the Court’s interpretation of the Constitution calls the contending sides of a national controversy to end their national division by accepting a common mandate rooted in the Constitution.

The Court is not asked to do this very often, having thus addressed the Nation only twice in our lifetime, in the decisions of Brown and Roe. But when the Court does act in this way, its decision requires an equally rare precedential force to counter the inevitable efforts to overturn it and to thwart its implementation. Some of those efforts may be mere unprincipled emotional reactions; others may proceed from principles worthy of profound respect. But whatever the premises of opposition may be, only the most convincing justification under accepted standards of precedent could suffice to demonstrate that a later decision overruling the first was anything but a surrender to political pressure, and an unjustified repudiation of the principle on which the Court staked its authority in the first instance. So to overrule under fire in the absence of the most compelling reason to reexamine a watershed decision would subvert the Court’s legitimacy beyond any serious question. Cf. Brown v. Board of Education, 349 U. S. 294, 300 (1955) (Broten II) (“[I]t should go without saying that the vitality of th[e] constitutional principles [announced in Brown I,] cannot be allowed to yield simply because of disagreement with them”).

The country’s loss of confidence in the Judiciary would be underscored by an equally certain and equally reasonable condemnation for another failing in overruling unnecessarily and under pressure. Some cost will be paid by anyone who approves or implements a constitutional decision where it is unpopular, or who refuses to work to undermine the decision or to force its reversal. The price may be criticism or ostracism, or it may be violence. An extra price will be paid by those who themselves disapprove of the decision’s results *868when viewed outside of constitutional terms, but who nevertheless struggle to accept it, because they respect the rule of law. To all those who will be so tested by following, the Court implicitly undertakes to remain steadfast, lest in the end a price be paid for nothing. The promise of constancy, once given, binds its maker for as long as the power to stand by the decision survives and the understanding of the issue has not changed so fundamentally as to render the commitment obsolete. From the obligation of this promise this Court cannot and should not assume any exemption when duty requires it to decide a case in conformance with the Constitution. A willing breach of it would be nothing, less than a breach of faith, and no Court that broke its faith with the people could sensibly expect credit for principle in the decision by which it did that.

It is true that diminished legitimacy may be restored, but only slowly. Unlike the political branches, a Court thus weakened could not seek to regain its position with a new mandate from the voters, and even if the Court could somehow go to the polls, the loss of its principled character could not be retrieved by the casting of so many votes. Like the character of an individual, the legitimacy of the Court must be earned over time. So, indeed, must be the character of a Nation of people who aspire to live according to the rule of law. Their belief in themselves as such a people is not readily separable from their understanding of the Court invested with the authority to decide their constitutional cases and speak before all others for their constitutional ideals. If the Court’s legitimacy should be undermined, then, so would the country be in its very ability to see itself through its constitutional ideals. The Court’s concern with legitimacy is not for the sake of the Court, but for the sake of the Nation to which it is responsible.

The Court’s duty in the present cases is clear. In 1973, it confronted the already-divisive issue of governmental power *869to limit personal choice to undergo abortion, for which it provided a new resolution based on the due process guaranteed by the Fourteenth Amendment. Whether or not a new social consensus is developing on that issue, its divisiveness is no less today than in 1973, and pressure to overrule the decision, like pressure to retain it, has grown only more intense. A decision to overrule Roe’s essential holding under the existing circumstances would address error, if error there was, at the cost of both profound and unnecessary damage to the Court’s legitimacy, and to the Nation’s commitment to the rule of law. It is therefore imperative to adhere to the essence of Roe’s original decision, and we do so today.

IV

From what we have said so far it follows that it is a constitutional liberty of the woman to have some freedom to terminate her pregnancy. We conclude that the basic decision in Roe was based on a constitutional analysis which we cannot now repudiate. The woman’s liberty is not so unlimited, however, that from the outset the State cannot show its concern for the life of the unborn, and at a later point in fetal development the State’s interest in life has sufficient force so that the right of the woman to terminate the pregnancy can be restricted.

That brings us, of course, to the point where much criticism has been directed at Roe, a criticism that always inheres when the Court draws a specific rule from what in the Constitution is but a general standard. We conclude, however, that the urgent claims of the woman to retain the ultimate control over her destiny and her body, claims implicit in the meaning of liberty, require us to perform that function. Liberty must not be extinguished for want of a line that is clear. And it falls to us to give some real substance to the woman’s liberty to determine whether to carry her pregnancy to full term.

*870We conclude the line should be drawn at viability, so that before that time the woman has a right to choose to terminate her pregnancy. We adhere to this principle for two reasons. First, as we have said, is the doctrine of stare deci-sis. Any judicial act of line-drawing may seem somewhat arbitrary, but Roe was a reasoned statement, elaborated with great care. We have twice reaffirmed it in the face of great opposition. See Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 759; Akron I, 462 U. S., at 419-420. Although we must overrule those parts of Thornburgh and Akron I which, in our view, are inconsistent with iBoe’s statement that the State has a legitimate interest in promoting the life or potential life of the unborn, see infra, at 882-883, the central premise of those cases represents an unbroken commitment by this Court to the essential holding of Roe. It is that premise which we reaffirm today.

The second reason is that the concept of viability, as we noted in Roe, is the time at which there is a realistic possibility of maintaining and nourishing a life outside the womb, so that the independent existence of the second life can in reason and all fairness be the object of state protection that now overrides the rights of the woman. See Roe v. Wade, 410 U. S., at 163. Consistent with other constitutional norms, legislatures may draw lines which appear arbitrary without the necessity of offering a justification. But courts may not. We must justify the lines we draw And there is no line other than viability which is more workable. To be sure, as we have said, there may be some medical developments that affect the precise point of viability, see supra, at 860, but this is an imprecision within tolerable limits given that the medical community and all those who must apply its discoveries will continue to explore the matter. The viability line also has, as a practical matter, an element of fairness. In some broad sense it might be said that a woman who fails to act before viability has consented to the State’s intervention on behalf of the developing child.

*871The woman’s right to terminate her pregnancy before viability is the most central principle of Roe v. Wade. It is a rule of law and a component of liberty we cannot renounce.

On the other side of the equation is the interest of the State in the protection of potential life. The Roe Court recognized the State’s “important and legitimate interest in protecting the potentiality of human life.” Roe, supra, at 162. The weight to be given this state interest, not the strength of the woman’s interest, was the difficult question faced in Roe. We do not need to say whether each of us, had we been Members of the Court when the valuation of the state interest came before it as an original matter, would have concluded, as the Roe Court did, that its weight is insufficient to justify a ban on abortions prior to viability even when it is subject to certain exceptions. The matter is not before us in the first instance, and coming as it does after nearly 20 years of litigation in Roe’s wake we are satisfied that the immediate question is not the soundness of Roe’s resolution of the issue, but the precedential force that must be accorded to its holding. And we have concluded that the essential holding of Roe should be reaffirmed.

Yet it must be remembered that Roe v. Wade speaks with clarity in establishing not only the woman’s liberty but also the State’s “important and legitimate interest in potential life.” Roe, supra, at 163. That portion of the decision in Roe has been given too little acknowledgment and implementation by the Court in its subsequent cases. Those eases decided that any regulation touching upon the abortion decision must survive strict scrutiny, to be sustained only if drawn in narrow terms to further a compelling state interest. See, e. g., Akron I, supra, at 427. Not all of the cases decided under that formulation can be reconciled with the holding in Roe itself that the State has legitimate interests in the health of the woman and in protecting the potential life within her. In resolving this tension, we choose to rely upon Roe, as against the later cases.

*872Roe established a trimester framework to govern abortion regulations. Under this elaborate but rigid construct, almost no regulation at all is permitted during the first trimester of pregnancy; regulations designed to protect the woman’s health, but not to further the State’s interest in potential life, are permitted during the second trimester; and during the third trimester, when the fetus is viable, prohibitions are permitted provided the life or health of the mother is not at stake. Roe, supra, at 163-166. Most of our cases since Roe have involved the application of rules derived from the trimester framework. See, e. g., Thornburgh v. American College of Obstetricians and Gynecologists, supra; Akron I, supra.

The trimester framework no doubt was erected to ensure that the woman’s right to choose not become so subordinate to the State’s interest in promoting fetal life that her choice exists in theory but not in fact. We do not agree, however, that the trimester approach is necessary to accomplish this objective. A framework of this rigidity was unnecessary and in its later interpretation sometimes contradicted the State’s permissible exercise of its powers.

Though the woman has a right to choose to terminate or continue her pregnancy before viability, it does not at all follow that the State is prohibited from taking steps to ensure that this choice is thoughtful and informed. Even in the earliest stages of pregnancy, the State may enact rules and regulations designed to encourage her to know that there are philosophic and social arguments of great weight that can be brought to bear in favor of continuing the pregnancy to full term and'that there are procedures and institutions to allow adoption of unwanted children as well as a certain degree of state assistance if the mother chooses to raise the child herself. “ ‘[T]he Constitution does not forbid a State or city, pursuant to democratic processes, from expressing a preference for normal childbirth.’” Webster v. Reproductive Health Services, 492 U. S., at 511 (opinion of *873the Court) (quoting Poelker v. Doe, 432 U. S. 519, 521 (1977)). It follows that States are free to enact laws to provide a reasonable framework for a woman to make a decision that has such profound and lasting meaning. This, too, we find consistent with Roe’s central premises, and indeed the inevitable consequence of our holding that the State has an interest in protecting the life of the unborn. .

We reject the trimester framework, which we do not consider to be part of the essential holding of Roe. See Webster v. Reproductive Health Services, 492 U. S., at 518 (opinion of Rehnquist, C. J.); id., at 529 (O’Connor, J., concurring in part and concurring in judgment) (describing the trimester framework as “problematic”). Measures aimed at ensuring that a woman’s choice contemplates the consequences for the ' fetus do not necessarily interfere with the right recognized in Roe, although those measures have been found to be inconsistent with the rigid trimester framework announced in that case. A logical reading of the central holding in Roe itself, and a necessary reconciliation of the liberty of the woman and the interest of the State in promoting prenatal life, require, in our view, that we abandon the trimester framework as a rigid prohibition on all previability regulation aimed at the protection of fetal life. The trimester framework suffers from these basic flaws: in its formulation it misconceives the nature of the pregnant woman’s interest; and in practice it undervalues the State’s interest in potential life, as recognized in Roe.

As our jurisprudence relating to all liberties save perhaps abortion has recognized, not every law which makes a right more difficult to exercise is, ipso facto, an infringement of that right. An example clarifies the point. We have held that not every ballot access limitation amounts to an infringement of the right to vote. Rather, the States are granted substantial flexibility in establishing the framework within which voters choose the candidates for whom' they *874wish to vote. Anderson v. Celebrezze, 460 U. S. 780, 788 (1983); Norman v. Reed, 502 U. S. 279 (1992).

The abortion right is similar. Numerous forms of state regulation might have the incidental effect of increasing the cost or decreasing the availability of medical care, whether for abortion or any other medical procedure. The fact that a law which serves a valid purpose, one not designed to strike at the right itself, has the incidental effect of making it more difficult or more expensive to procure an abortion cannot be enough to invalidate it. Only where state regulation imposes an undue burden on a woman’s ability to make this decision does the power of the State reach into the heart of the liberty protected by the Due Process Clause. See Hodgson v. Minnesota, 497 U. S. 417, 458-459 (1990) (O’Connor, J., concurring in part and concurring in judgment in part); Ohio v. Akron Center for Reproductive Health, 497 U. S. 502, 519-520 (1990) (Akron II) (opinion of Kennedy, J.); Webster v. Reproductive Health Services, supra, at 530 (O’Connor, J., concurring in part and concurring in judgment); Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 828 (O’Connor, J., dissenting); Simopoulos v. Virginia, 462 U. S. 506, 520 (1983) (O’Connor, J., concurring in part and concurring in judgment); Planned Parenthood Assn. of Kansas City, Mo., Inc. v. Ashcroft, 462 U. S. 476, 505 (1983) (O’Connor, J., concurring in judgment in part and dissenting in part); Akron I, 462 U. S., at 464 (O’Connor, J., joined by White and Rehnquist, JJ., dissenting); Bellotti v. Baird, 428 U. S. 132, 147 (1976) (Bellotti I).

For the most part, the Court’s early abortion cases adhered to this view. In Maher v. Roe, 432 U. S. 464, 473-474 (1977), the Court explained: “Roe did not declare an unqualified ‘constitutional right to an abortion,’ as the District Court seemed to think. Rather, the right protects the woman from unduly burdensome interference with her freedom to decide whether to terminate her pregnancy.” See *875also Doe v. Bolton, 410 U. S. 179, 198 (1973) (“[T]he interposition of the hospital abortion committee is unduly restrictive of the patient’s rights”); Bellotti I, supra, at 147 (State may not “impose undue burdens upon a minor capable of giving an informed consent”); Harris v. McRae, 448 U. S. 297, 314 (1980) (citing Maker, supra). Cf. Carey v. Population Services International, 431 U. S., at 688 (“[T]he same test must be applied to state regulations that burden an individual’s right to decide to prevent conception or terminate pregnancy by substantially limiting access to the means of effectuating that decision as is applied to state statutes that prohibit the decision entirely”).

These considerations of the nature of the abortion right illustrate that it is an overstatement to describe it as a right to decide whether to have an abortion “without interference from the State.” Planned Parenthood of Central Mo. v. Danforth, 428 U. S. 52, 61 (1976). All abortion regulations interfere to some degree with a woman’s ability to decide whether to terminate her pregnancy. It is, as a consequence, not surprising that despite the protestations contained in the original Roe opinion to the effect that the Court was not recognizing an absolute right, 410 U. S., at 154-155, the Court’s experience applying the trimester framework has led to the striking down of some abortion regulations which in no real sense deprived women of the ultimate decision. Those decisions went too far because the right recognized by Roe is a right “to be free from unwarranted governmental intrusion into matters so fundamentally affecting a person as the decision whether to bear or beget a child.” Eisenstadt v. Baird, 405 U. S., at 453. Not all governmental intrusion is of necessity unwarranted; and that brings us to the other basic flaw in the trimester framework: even in ftoe’s terms, in practice it undervalues the State’s interest in the potential life within the woman.

Roe v. Wade was express in its recognition of the State’s “important and legitimate interests] in preserving and pro-*876teeting the health of the pregnant woman [and] in protecting the potentiality of human life.” 410 U. S., at 162. The trimester framework, however, does not fulfill Roe’s own promise that the State has an interest in protecting fetal life or potential life. Roe began the contradiction by using the trimester framework to forbid any regulation of abortion designed to advance that interest before viability. Id., at 163. Before viability, Roe and subsequent cases treat all governmental attempts to influence a woman’s decision on behalf of the potential life within her as unwarranted. This treatment is, in our judgment, incompatible with the recognition that there is a substantial state interest in potential life throughout pregnancy. Cf. Webster, 492 U. S., at 519 (opinion of Rehnquist, C. J.); Akron I, supra, at 461 (O’Connor, J., dissenting).

The very notion that the State has a substantial interest in potential life leads to the conclusion that not all regulations must be deemed unwarranted. Not all burdens on the right to decide whether to terminate a pregnancy will be undue. In our view, the undue burden standard is the appropriate means of reconciling the State’s interest with the' woman’s constitutionally protected liberty.

The concept of an undue burden has been utilized by the Court as well as individual Members of the Court, including two of us, in ways that could be considered inconsistent. See, e. g., Hodgson v. Minnesota, supra, at 459-461 (O’CONNOR, J., concurring in part and concurring in judgment); Akron II, supra, at 519-520 (opinion of Kennedy, J.); Thornburgh v. American College of Obstetricians and Gynecologists, supra, at 828-829 (O’Connor, J., dissenting); Akron I, supra, at 461-466 (O’Connor, J., dissenting); Harris v. McRae, supra, at 314; Maher v. Roe, supra, at 473; Beal v. Doe, 432 U. S. 438, 446 (1977); Bellotti I, supra, at 147. Because we set forth a standard of general application to which we intend to adhere, it is important to clarify what is meant by an undue burden.

*877A finding of an undue burden is a shorthand for the conclusion that a state regulation has the purpose or effect of placing a substantial obstacle in the path of a woman seeking an abortion of a nonviable fetus. A statute with this purpose is invalid because the means chosen by the State to further the interest in potential life must be calculated to inform the woman’s free choice, not hinder it. And a statute which, while furthering the interest in potential life or some other valid state interest, has the effect of placing a substantial obstacle in the path of a woman’s choice cannot be considered a permissible means of sérving its legitimate ends. To the extent that the opinions of the Court or of individual Justices use the undue burden standard in a manner that is inconsistent with this analysis, we set out what in our view should be the controlling standard. Cf. McCleskey v. Zant, 499 U. S. 467, 489 (1991) (attempting “to define the doctrine of abuse of the writ with more precision” after acknowledging tension among earlier cases). In our considered judgment, an undue burden is an unconstitutional burden. See Akron II, 497 U. S., at 519-520 (opinion of Kennedy, J.). Understood another way, we answer the question, left open in previous opinions discussing the undue burden formulation, whether a law designed to further the State’s interest in fetal life which imposes an undue burden on the woman’s decision before fetal viability could be constitutional. See, e. g., Akron I, 462 U. S., at 462-463 (O’Connor, J., dissenting). The answer is no.

Some guiding principles should emerge. What is at stake is the woman’s right to make the ultimate decision, not a right to be insulated from all others in doing so. Regulations which do no more than create a structural mechanism by which the State, or the parent or guardian of a minor, may express profound respect for the life of the unborn are permitted, if they are not a substantial obstacle to the woman’s exereise of the right to choose. See infra, at 899-900 (addressing Pennsylvania’s parental consent requirement). *878Unless it has that effect on her right of choice, a state measure designed to persuade her to choose childbirth over abortion will be upheld if reasonably related to that goal.' Regulations designed to foster the health of a woman seeking an abortion are valid if they do not constitute an undue burden.

Even when jurists reason from shared premises, some disagreement is inevitable. Compare Hodgson, 497 U. S., at 482-497 (Kennedy, J., concurring in judgment in part and dissenting in part), with id., at 458-460 (O’Connor, J., concurring in part and concurring in judgment in part). That is to be expected in the application of any legal standard which must accommodate life’s complexity. We do not expect it to be otherwise with respect to the undue burden standard. We give this summary:

(a) To protect the central right recognized by Roe v. Wade while at the same time accommodating the State’s profound interest in potential life, we will employ the undue burden analysis as explained in this opinion. An undue burden exists, and therefore a provision of law is invalid, if its purpose or effect is to place a substantial obstacle in the path of a woman seeking an abortion before the fetus attains viability.

(b) We reject the rigid trimester framework of Roe v. Wade. To promote the State’s profound interest in potential life, throughout pregnancy the State may take measures to ensure that the woman’s choice is informed, and measures designed to advance this interest will not be invalidated as long as their purpose is to persuade the woman to choose childbirth over abortion. These measures must not be an undue burden on the right.

(c) As with any medical procedure, the State may enact regulations to further the health or safety of a woman seeking an abortion. Unnecessary health regulations that have the purpose or effect of presenting a substantial obstacle to a woman seeking an abortion impose an undue burden on the right.

*879(d) Our adoption of the undue burden analysis does not disturb the central holding of Roe v. Wade, and we reaffirm that holding. Regardless of whether exceptions are made for particular circumstances, a State may not prohibit any woman from making the ultimate decision to terminate her pregnancy before viability.

(e) We also reaffirm Roe’s holding that “subsequent to viability, the State in promoting its interest in the potentiality of human life may, if it chooses, regulate, and even proscribe, abortion except where it is necessary, in appropriate medical judgment, for the preservation of the life or health of the mother.” Roe v. Wade, 410 U. S., at 164-165.

These principles control our assessment of the Pennsylvania statute, and we now turn to the issue of the validity of its challenged provisions.

V

The Court of Appeals applied what it believed to be the undue burden standard and upheld each' of the provisions except for the husband notification requirement. We agree generally with this conclusion, but refine the undue burden analysis in accordance with the principles articulated above. We now consider the separate statutory sections at issue.

A

Because it is central to the operation of various other requirements, we begin with the statute’s definition of medical emergency. Under the statute, a medical emergency is

“[t]hat condition which, on the basis of the physician’s good faith clinical judgment, so complicates the medical condition of a pregnant woman as to necessitate the immediate abortion of her pregnancy to avert her death or for which a delay will create serious risk of substantial and irreversible impairment of a major bodily function.” 18 Pa. Cons. Stat. §3208 (1990).

*880Petitioners argue that the definition is too narrow, contending that it forecloses the possibility of an immediate abortion despite some significant health risks. If the contention were correct, we would be required to invalidate the restrictive operation of the provision, for the essential holding of Roe forbids a State to interfere with a woman’s choice to undergo an abortion procedure if continuing her pregnancy would constitute a threat to her health. 410 U. S., at 164. See also Harris v. McRae, 448 U. S., at 316.

The District Court found that there were three serious conditions which would not be covered by the statute: pre-eclampsia, inevitable abortion, and premature ruptured membrane. 744 F. Supp., at 1378. Yet, as the Court of Appeals observed, 947 F. 2d, at 700-701, it is undisputed that under some circumstances each of these conditions could lead to an illness with substantial and irreversible consequences. While the definition could be interpreted in an unconstitutional manner, the Court of Appeals construed the phrase “serious risk” to include those circumstances. Id., at 701. It stated: “[W]e read the medical emergency exception as intended by the Pennsylvania legislature to assure that compliance with its abortion regulations would not in any way pose a significant threat to the life or health of a woman.” Ibid. As we said in Brockett v. Spokane Arcades, Inc., 472 U. S. 491, 499-500 (1985): “Normally, ... we defer to the construction of a state statute given it by the lower federal courts.” Indeed, we have said that we will defer to lower court interpretations of state law unless they amount to “plain” error. Palmer v. Hoffman, 318 U. S. 109, 118 (1943). This “ ‘reflects] our belief that district courts and courts of appeals are better schooled in and more able to interpret the laws of their respective States.’” Frisby v. Schultz, 487 U. S. 474, 482 (1988) (citation omitted). We adhere to that course today, and conclude that, as construed by the Court of Appeals, the medical emergency definition imposes no undue burden on a woman’s abortion right.

*881B

We next consider the informed consent requirement. 18 Pa. Cons. Stat. §3205 (1990). Except in a medical emergency, the statute requires that at least 24 hours before performing an abortion a physician inform the woman of the nature of the procedure, the health risks of the abortion and of childbirth, and the “probable gestational age of the unborn child.” The physician or a qualified nonphysician must inform the woman of the availability of printed materials published by the State describing the fetus and providing information about medical assistance for childbirth, information about child support from the father, and a list of agencies which provide adoption and other services as alternatives. to abortion. An abortion may not be performed unless the woman certifies in writing that she has been informed of the availability of these printed materials and has been provided them if she chooses to view them.

Our prior decisions establish that as with any medical procedure, the State may require a woman to give her written informed consent to an abortion. See Planned Parenthood of Central Mo. v. Danforth, 428 U. S., at 67. In this respect, the statute is unexceptional. Petitioners challenge the statute’s definition of informed consent because it includes the provision of specific information by the doctor and the mandatory 24-hour waiting period. The conclusions reached by a majority of the Justices in the separate opinions filed today and the undue burden standard adopted in this opinion require us to overrule in part some of the Court’s past decisions, decisions driven by the trimester framework’s prohibition of all previability regulations designed to further the State’s interest in fetal life.

In Akron I, 462 U. S. 416 (1983), we invalidated an ordinance which required that a woman seeking an abortion be provided by her physieian with specific information “designed. to influence the woman’s informed choice between abortion or childbirth.” Id., at 444. As we later described *882the Akron I holding in Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 762, there were two purported flaws in the Akron ordinance: the information was designed to dissuade the woman from having an abortion and the ordinance imposed “a rigid requirement that a specific body of information be given in all cases, irrespective of the particular needs of the patient. . . .” Ibid.

To the extent Akron I and Thornburgh find a constitutional violation when the government requires, as it does here, the giving of truthful, nonmisleading information about the nature of the procedure, the attendant health risks and those of childbirth, and the “probable gestational age” of the fetus, those eases go too far, are inconsistent with Roe’s acknowledgment of an important interest in potential life, and are overruled. This is clear even on the very terms of Akron I and Thornburgh. Those decisions, along with Dan-forth, recognize a substantial government interest justifying a requirement that a woman be apprised of the health risks of abortion and childbirth. E. g., Danforth, supra, at 66-67. It cannot be questioned that psychological well-being is a facet of health. Nor can it be doubted that most women considering an abortion would deem the impact on the fetus relevant, if not dispositive, to the decision. In attempting to ensure that a woman apprehend the full consequences of her decision, the State furthers the legitimate purpose of reducing the risk that a woman may elect an abortion, only to discover later, with devastating psychological consequences, that her decision was not fully informed. If the information the State requires to be made available to the woman is truthful and not misleading, the requirement may be permissible.

We also see no reason why the State may not require doctors to inform a woman seeking an abortion of the availability of materials relating to the consequences to the fetus, even when those consequences have no direct relation to her health. An example illustrates the point. We would think *883it constitutional for the State to require that in order for there to he informed consent to a kidney transplant operation the recipient must be supplied with information about risks to the donor as well as risks to himself or herself. A requirement that the physician make available information similar to that mandated by the statute here was described in Thornburgh as “an outright attempt to wedge the Commonwealth’s message discouraging abortion into the privacy of the informed-consent dialogue between the woman and her physician.” 476 U. S., at 762. We conclude, however, that informed choice need not be defined in such narrow terms that all considerations of the effect on the fetus are made irrelevant. As we have made clear, we depart from the holdings of Akron I and Thornburgh to the extent that we permit a State to further its legitimate goal of protecting the life of the unborn by enacting legislation aimed at ensuring a decision that is mature and informed, even when in so doing the State expresses a preference for childbirth over abortion. In short, requiring that the woman be informed of the availability of information relating to fetal development and the assistance available should she decide to carry the pregnancy to full term is a reasonable measure to ensure an informed choice, one which might cause the woman to choose childbirth over abortion. This requirement cannot be considered a substantial obstacle to obtaining an abortion, and, it follows, there is no undue burden.

Our prior cases also suggest that the “straitjacket,” Thornburgh, supra, at 762 (quoting Danforth, supra, at 67, n. 8), of particular information which must be given in each ease interferes with a constitutional right of privacy between a pregnant woman and her physician. As a preliminary matter, it is worth noting that the statute now before us does not require a physician to comply with the informed consent provisions “if he or she can demonstrate by a preponderance of the evidence, that he or she reasonably believed that furnishing the information would have resulted in a severely *884adverse effect on the physical or mental health of the patient/’ 18 Pa. Cons. Stat. § 3205 (1990). In this respect, the statute does not prevent the physician from exercising his or her medical judgment.

Whatever constitutional status the doctor-patient relation may have as a general matter, in the present context it is derivative of the woman’s position. The doctor-patient relation does not underlie or override the two more general rights under which the abortion right is justified: the right to make family decisions and the right to physical autonomy. On its own, the doctor-patient relation here is entitled to the same solicitude it receives in other contexts. Thus, a requirement that a doctor give a woman certain information as part of obtaining her consent to an abortion is, for constitutional purposes, no different from a requirement that a doctor give certain specific'information about any medical procedure.

All that is left of petitioners’ argument is an asserted First Amendment right of a physician not to provide information about the risks of abortion, and childbirth, in a manner mandated by the State. To be sure, the physician’s First Amendment rights not to speak are implicated, see Wooley v. Maynard, 430 U. S. 705 (1977), but only as part of the practice of medicine, subject to reasonable licensing and regulation by the State, cf. Whalen v. Roe, 429 U. S. 589, 603 (1977). We see no constitutional infirmity in the requirement that the physician provide the information mandated by the State here.

The Pennsylvania statute also requires us to reconsider the holding in Akron I that the State may not require that a physician, as opposed to a qualified assistant, provide information relevant to a woman’s informed consent. 462 U. S., at 448. Since there is no evidence on this record that requiring a doctor to give the information as provided by the statute would amount in practical terms to a substantial obstacle to a woman seeking an abortion, we conclude that it is not *885an undue burden. Our eases reflect the fact that the Constitution gives the States broad latitude to decide that particular functions may be performed only by licensed professionals, even if an objective assessment might suggest that those same tasks could be performed by others. See Williamson v. Lee Optical of Okla., Inc., 348 U. S. 483 (1955). Thus, we uphold the provision as a reasonable means to ensure that the woman’s consent is informed.

Our analysis of Pennsylvania’s 24-hour waiting period between the provision of the information deemed necessary to informed consent and the performance of an abortion under the undue burden standard requires us to reconsider the premise behind the decision in Akron I invalidating a parallel requirement. In Akron I we said: “Nor are we convinced that the State’s legitimate concern that the woman’s decision be informed is reasonably served by requiring a 24-hour delay as a matter of course.” 462 U. S., at 450. We consider that conclusion to be wrong. The idea that important decisions will be more informed and deliberate if they follow some period of reflection does not strike us as unreasonable, particularly where the statute directs that important information become part of the background of the decision. The statute, as construed by the Court of Appeals, permits avoidance of the waiting period in the event of a medical emergency and the record evidence shows that in the vast majority of cases, a 24-hour delay does not create any appreciable health risk. In theory, at least, the waiting period is a reasonable measure to implement the State’s interest in protecting the life of the unborn, a measure that does not amount to an undue burden.

Whether the mandatory 24-hour waiting period is nonetheless invalid because in practice it is a substantial obstacle to a woman’s choice to terminate her pregnancy is a closer question. The findings of fact by the District Court indicate that because of the distances many women must travel to reach an abortion provider, the practical effect will often be *886a delay of much more than a day because the waiting period requires that a woman seeking an abortion make at least two visits to the doctor. The District Court also found that in many instances this will increase the exposure of women seeking abortions to “the harassment and hostility of antiabortion protestors demonstrating outside a clinic.” 744 F. Supp., at 1351. As a result, the District Court found that for those women who have the fewest financial resources, those who must travel long distances, and those who have difficulty explaining their whereabouts to husbands, employers, or others, the 24-hour waiting period will be “particularly burdensome.” Id., at 1352.

These findings are troubling in some respects, but they do not demonstrate that the waiting period constitutes an undue burden. We do not doubt that, as the District Court held, the waiting period has the effect of “increasing the cost and risk of delay of abortions,” id., at 1378, but the District Court did not conclude that the increased costs and potential delays amount to substantial obstacles. Rather, applying the trimester framework’s strict prohibition of all regulation designed to promote the State’s interest in potential life before viability, see id., at 1374, the District Court concluded that the waiting period does not further the state “interest in maternal health” and “infringes the physician’s discretion to exercise sound medical judgment,” id., at 1378. Yet, as we have stated, under the undue burden standard a State is permitted to enact persuasive measures which favor childbirth over abortion, even if those measures do not further a health interest. And while the waiting period does limit a physician’s discretion, that is not, standing alone, a reason to invalidate it. In light of the construction given the statute’s definition of medical emergency by the Court of Appeals, and the District Court’s findings, we cannot say that the waiting period imposes a real health risk.

We also disagree with the District Court’s conclusion that the “particularly burdensome” effects of the waiting period *887on some women require its invalidation. A particular burden is not of necessity a substantial obstacle. Whether a burden falls on a particular group is a distinct inquiry from whether it is a substantial obstacle even as to the women in that group. And the District Court did not conclude that the waiting period is such an obstacle even for the women who are most burdened by it. Hence, on the record before us, and in the context of this facial challenge, we are not convinced that the 24-hour waiting period constitutes an undue burden.

We are left with the argument that the various aspects of the informed consent requirement are unconstitutional because they place barriers in the way of abortion on demand. Even the broadest reading of Roe, however, has not suggested that there is a constitutional right to abortion on demand. See, e. g., Doe v. Bolton, 410 U. S., at 189. Rather, the right protected by Roe is a right to decide to terminate a pregnancy free of undue interference by the State. Because the informed consent requirement facilitates the wise exercise of that right, it cannot be classified as an interference with the right Roe protects. The informed consent requirement is not an undue burden on that right.

C

Section 8209 of Pennsylvania’s abortion law provides, except in eases of medical emergency, that no physician shall perform an abortion on a married woman without receiving a signed statement from the woman that she has notified her spouse that she is about to undergo an abortion. The woman has the option of providing an alternative signed statement certifying that her husband is not the man who impregnated her; that her husband could not be located; that the pregnancy is the result of spousal sexual assault which she has reported; or that the woman believes that notifying her husband will cause him or someone else to inflict bodily injury upon her. A physician who performs an abortion on *888a married woman without receiving the appropriate signed statement will have his or her license revoked, and is liable to the husband for damages.

The District Court heard the testimony of numerous expert witnesses, and made detailed findings of fact regarding the effect of this statute. These included:

“273. The vast majority of women consult their husbands prior to deciding to terminate their pregnancy....

“279. The ‘bodily injury’ exception could not be invoked by a married woman whose husband, if notified, would, in her reasonable belief, threaten to (a) publicize her intent to have an abortion to family, friends or acquaintances; (b) retaliate against her in future child custody or divorce proceedings; (e) inflict psychological intimidation or emotional harm upon her, her children or other persons; (d) inflict bodily harm on other persons such as children, family members or other loved ones; or (e) use his control over finances to deprive of necessary monies for herself or her children....

“281. Studies reveal that family violence occurs in two million families in the United States.. This figure, however, is a conservative one that substantially understates (because battering is usually not reported until it reaches life-threatening proportions) the actual number of families affected by domestic violence. In fact, researchers estimate that one of every two women will be battered at some time in their life....

“282. A wife may not elect to notify her husband of her intention to have an abortion for a variety of reasons, including the husband’s illness, concern about her own health, the imminent failure of the marriage, or the husband’s absolute opposition to the abortion. ...

“283. The required filing of the spousal consent form would require plaintiff-clinics to change their counseling *889procedures and force women to reveal their most intimate decision-making on pain of criminal sanctions. The confidentiality of these revelations could not be guaranteed, since the woman’s records are not immune from subpoena....

“284. Women of all class levels, educational backgrounds, and racial, ethnic and religious groups are battered....

“285. Wife-battering or abuse can take on many physical and psychological forms. The nature and scope of the battering can cover a broad range of actions and be gruesome and torturous....

“286. Married women, victims of battering, have been killed in Pennsylvania and throughout the United States....

“287. Battering can often involve a substantial amount of sexual abuse, including marital rape and sexual mutilation....

“288. In a domestic abuse situation, it is common for the battering husband to also abuse the children in an attempt to coerce the wife....

“289. Mere notification of pregnancy is frequently a flashpoint for battering and violence within the family. The number of battering incidents is high during the pregnancy and often the worst abuse can be associated with pregnancy. . . . The battering husband may deny parentage and use the pregnancy as an excuse for abuse....

“290. Secrecy typically shrouds abusive families. Family members are instructed not to tell anyone, especially police or doctors, about the abuse and violence. Battering husbands often threaten their wives or her children with further abuse if she tells an outsider of the violence and tells her that nobody will believe her. A battered woman, therefore, is highly unlikely to disclose *890the violence against her for fear of retaliation by the abuser....

“291. Even when confronted directly by medical personnel or other helping professionals, battered women often will not admit to the battering because they have not admitted to themselves that they are battered.. . .

“294. A woman in a shelter or a safe house unknown to her husband is not ‘reasonably likely’ to have bodily harm inflicted upon her by her batterer, however her attempt to notify her husband pursuant to section 3209 could accidentally disclose her whereabouts to her husband. Her fear of future ramifications would be realistic under, the circumstances.

“295. Marital rape is rarely discussed with others or reported to law enforcement authorities, and of those reported only few are prosecuted....

“296. It is common for battered women to have sexual intercourse with their husbands to avoid being battered. While this type of coercive sexual activity would be spousal sexual assault as defined by the Act, many women may not consider it to be so and others would fear disbelief....

“297. The marital rape exception to section 3209 cannot be claimed by women who are victims of coercive sexual behavior other than penetration. The 90-day reporting requirement of the spousal sexual assault statute, 18 Pa. Con. Stat. Ann. § 3218(c), further narrows the class of sexually abused wives who can claim the exception, since many of these women may be psychologically unable to discuss or report the rape for several years after the incident. . ..

“298. Because of the nature of the battering relationship, battered women are unlikely to avail themselves of the exceptions to section 3209 of the Act, regardless of *891whether the section applies to them.” 744 F. Supp., at 1360-1362 (footnote omitted).

These findings are supported by studies of domestic violence. The American Medical Association (AMA) has published a summary of the recent research in this field, which indicates that in an average 12-month period in this country, approximately two million women are the victims of severe assaults by their male partners. In a 1985 survey, women reported that nearly one of every eight husbands had assaulted their wives during the past year. The AMA views these figures as “marked underestimates,” because the nature of these incidents discourages women from reporting them, and because surveys typically exclude the very poor, those who do not speak English well, and women who are homeless or in institutions or hospitals when the survey is conducted. According to the AMA, “[researchers on family violence agree that the true incidence of partner violence is probably double the above estimates; or four million severely assaulted women per year. Studies on prevalence suggest that from one-fifth to one-third of all women will be physically assaulted by a partner or ex-partner during their lifetime.” AMA Council on Scientific Affairs, Violence Against Women 7 (1991) (emphasis in original). Thus on an average day in the United States, nearly 11,000 women are severely assaulted by their male partners. Many of these incidents involve sexual assault. Id., at 3-4; Shields & Hanneke, Battered Wives’ Reactions to Marital Rape, in The Dark Side of Families: Current Family Violence Research 131, 144 (D. Finkelhor, R. Gelles, G. Hataling, & M. Straus eds. 1983). In families where wifebeating takes. place, moreover, child abuse is often present as well. Violence Against Women, supra, at 12.

Other studies fill in the rest of this troubling picture. Physical violence is only the most visible form of abuse. Psychological abuse, particularly forced social and economic isolation of women, is also eommon. L. Walker, The Bat*892tered Woman Syndrome 27-28 (1984). Many victims of domestic violence remain with their abusers, perhaps because they perceive no superior alternative. Herbert, Silver, & Ellard, Coping with an Abusive Relationship: I. How and Why do Women Stay?, 53 J. Marriage & the Family 311 (1991). Many abused women who find temporary refuge in shelters return to their husbands, in large part because they have no other source of income. Aguirre, Why Do They Return? Abused Wives in Shelters, 30 J. Nat. Assn, of Social Workers 350, 352 (1985). Returning to one’s abuser can be dangerous. Recent Federal Bureau of Investigation statistics disclose that 8.8 percent of all homicide victims in the United States are killed by their spouses. Mercy & Saltz-man, Fatal Violence Among Spouses in the United States, 1976-85, 79 Am. J. Public Health 595 (1989). Thirty percent of female homicide victims are killed by their male partners. Domestic Violence: Terrorism in the Home, Hearing before the Subcommittee on Children, Family, Drugs and Alcoholism of the Senate Committee on Labor and Human Resources, 101st Cong., 2d Sess., 3 (1990).

The limited research that has been conducted with respect to notifying one’s husband about an abortion, although involving samples too small to be representative, also supports the District Court’s findings of fact. The vast majority of women notify their male partners of their decision to obtain an abortion. In many cases in which married women do not notify their husbands, the pregnancy is the result of an extramarital affair. Where the husband is the father, the primary reason women do not notify their husbands is that the husband and wife are experiencing marital difficulties, often accompanied by incidents of violence. Ryan & Plutzer, When Married Women Have Abortions: Spousal Notification and Marital Interaction, 51 J. Marriage & the Family 41, 44 (1989).

This information and the District Court’s findings reinforce what common sense would suggest. In well-*893functioning marriages, spouses discuss important intimate decisions such as whether to bear a child. But there are millions of women in this country who are the victims of regular physical and psychological abuse at the hands of their husbands. Should these women become pregnant, they may have very good reasons for not wishing to inform their husbands of their decision to obtain an abortion. Many may have justifiable fears of physical abuse, but may be no less fearful of the consequences of reporting prior abuse to the Commonwealth of Pennsylvania. Many may have a reasonable fear that notifying their husbands will provoke further instances of child abuse; these women are not exempt from § 3209’s notification requirement. Many may fear devastating forms of psychological abuse from their husbands, including verbal harassment, threats of future violence, the destruction of possessions, physical confinement to the home, the withdrawal of financial support, or the disclosure of the abortion to family and friends. These methods of psychological abuse may act as even more of a deterrent to notification than the possibility of physical violence, but women who are the victims of the abuse are not exempt from § 8209’s notifi-. cation requirement. And many women who are pregnant as a result of sexual assaults by their husbands will be unable to avail themselves of the exception for spousal sexual assault, § 3209(b)(3), because the exception requires that the woman have notified law enforcement authorities within 90 days of the assault, and her husband will be notified of her report once an investigation begins, § 3128(c). If anything in this field is certain, it is that victims of spousal sexual assault are extremely reluctant to report the abuse to the government; hence, a great many spousal rape victims will not be exempt from the notification requirement imposed by §3209.

The spousal notification requirement is thus likely to prevent a significant number of women from obtaining an abortion. It does not merely make abortions a little more difficult or expensive to obtain; for. many women, it will impose *894a substantial obstacle. We must not blind ourselves to the fact that the significant number of women who fear for their safety and the safety of their children are likely to be deterred from procuring an abortion as surely as if the Commonwealth had outlawed abortion in all cases.

Respondents attempt to avoid the conclusion that § 3209 is invalid by pointing out that it imposes almost no burden at all for the vast majority of women seeking abortions. They begin by noting that only about 20 percent of the women who obtain abortions are married. They then note that of these women about 95 percent notify their husbands of their own volition. Thus, respondents argue, the effects of § 3209 are felt by only one percent of the women who obtain abortions. Respondents argue that since some of these women will be able to notify their husbands without adverse consequences or will qualify for one of the exceptions, the statute affects fewer than one percent of women seeking abortions. For this reason, it is asserted, the statute cannot be invalid on its face. See Brief for Respondents 83-86. We disagree with respondents’ basic method of analysis.

The analysis does not end with the one percent of women upon whom the statute operates; it begins there. Legislation is measured for consistency with the Constitution by its impact on those whose conduct it affects. For example, we would not say that a law which requires a newspaper to print a candidate’s reply to an unfavorable editorial is valid on its face because most newspapers would adopt the policy even absent the law. See Miami Herald Publishing Co. v. Tornillo, 418 U. S. 241 (1974). The proper focus of constitutional inquiry is the group for whom the law is a restriction, not the group for whom the law is irrelevant.

Respondents’ argument itself gives implicit recognition to this principle, at one of its critical points. Respondents speak of the one percent of women seeking abortions who are married and would choose not to notify their husbands of their plans. By selecting as the controlling class women *895who wish to obtain abortions, rather than all women or all pregnant women, respondents in effect concede that §3209 must be judged by reference to those for whom it is an actual rather than an irrelevant restriction. Of course, as we have said, §3209’s real target is narrower even than the class of women seeking abortions identified by the State: it is married women seeking abortions who do not wish to notify their husbands of their intentions and who do not qualify for one of the statutory exceptions to the notice requirement. The unfortunate yet persisting conditions we document above will mean that in a large fraction of the cases in which §3209 is relevant, it will operate as a substantial obstacle to a woman’s choice to undergo an abortion. It is an undue burden, and therefore invalid.

This conclusion is in no way inconsistent with our decisions upholding parental notification or consent requirements. See, e. g., Akron II, 497 U. S., at 510-519; Bellotti v. Baird, 443 U. S. 622 (1979) (Bellotti II); Planned Parenthood of Central Mo. v. Danforth, 428 U. S., at 74. Those enactments, and our judgment that they are constitutional, are based on the quite reasonable assumption that minors will benefit from consultation with their parents and that children will often not realize that their parents have their best interests at heart. We cannot adopt a parallel assumption about adult women.

We recognize that a husband has a “deep and proper concern and interest ... in his wife’s pregnancy and in the growth and development of the fetus she is' carrying.” Dan-forth, supra, at 69. With regard to the children he has fathered and raised, the Court has recognized his “cognizable and substantial” interest in their custody. Stanley v. Illinois, 405 U. S. 645, 651-652 (1972); see also Quilloin v. Walcott, 434 U. S. 246 (1978); Caban v. Mohammed, 441 U. S. 380 (1979); Lehr v. Robertson, 463 U. S. 248 (1983). If these cases concerned a State’s ability to require the mother to notify the father before taking some action with respect to a living *896child raised by both, therefore, it would be reasonable to conclude as a general matter that the father’s interest in the welfare of the child and the mother’s interest are equal.

Before birth, however, the issue takes on a very different cast. It is an inescapable biological fact that state regulation with respect to the child a woman is carrying will have a far greater impact on the mother’s liberty than on the father’s. The effect of state regulation on a woman’s protected liberty is doubly deserving of scrutiny in such a case, as the State has touched not only upon the private sphere of the family but upon the very bodily integrity of the pregnant woman. Cf. Cruzan v. Director, Mo. Dept. of Health, 497 U. S., at 281. The Court has held that “when the wife and the husband disagree on this decision, the view of only one of the two marriage partners can prevail. Inasmuch as it is the woman who physically bears the child and who is the more directly and immediately affected by the pregnancy, as between the two, the balance weighs in her favor.” Danforth, supra, at 71. This conclusion rests upon the basic nature of marriage and the nature of our Constitution: “[T]he marital couple is not an independent entity with a mind and heart of its own, but an association of two individuals each with a separate intellectual and emotional makeup. If the right of privacy means anything, it is the right of the individual, married or single, to be free from unwarranted governmental intrusion into matters so fundamentally affecting a person as the decision whether to bear or beget a child.” Eisenstadt v. Baird, 405 U. S., at 453 (emphasis in original). The Constitution protects individuals, men and women alike, from unjustified state interference, even when that interference is enacted into law for the benefit of their spouses.

There was a time, not so long ago, when a different understanding of the family and of the Constitution prevailed. In Bradwell v. State, 16 Wall. 130 (1873), three Members of this *897Court reaffirmed the common-law principle that “a woman had no legal existence separate from her husband, who was regarded as her head and representative in the social state; and, notwithstanding some recent modifications of this civil status, many of the special rules of law flowing from and dependent upon this cardinal principle still exist in foil force in most States.” Id., at 141 (Bradley, X, joined by Swayne and Field, JX, concurring in judgment). Only one generation has passed since this Court observed that “woman is still regarded as the center of home and family life,” with attendant “special responsibilities” that precluded full and independent legal status under the Constitution. Hoyt v. Florida, 368 U. S. 57, 62 (1961). These views, of course, are no longer consistent with our understanding of the family, the individual, or the Constitution.

In keeping with our rejection of the common-law understanding of a woman’s role within the family, the Court held in Danforth that the Constitution does not permit a State to require a married woman to obtain her husband’s consent before undergoing an abortion. 428 U. S., at 69. The principles that guided the Court in Danforth should be our guides today. For the great many women who are victims of abuse inflicted by their husbands, or whose children are the victims of such abuse, a spousal notice requirement enables the husband to wield an effective veto over his wife’s decision. Whether the prospect of notification itself deters such women from seeking abortions, or whether the husband, through physical force or psychological pressure or economic coercion, prevents his wife from obtaining an abortion until it is too late, the notice requirement will often be tantamount to the veto found unconstitutional in Danforth. The women most affected by this.law — those who most reasonably fear the consequences of notifying their husbands that they are pregnant — are in the gravest danger.

*898The husband’s interest in the life of the child his wife is carrying does not permit the State to empower him with this troubling degree of authority over his wife. The contrary view leads to consequences reminiscent of the common law. A husband has no enforceable right to require a wife to advise him before she exercises her personal choices. If a husband’s interest in the potential life of the child outweighs a wife’s liberty, the State could require a married woman to notify her husband before she uses a postfertilization contraceptive. Perhaps next in line would be a statute requiring pregnant married women to notify their husbands before engaging in conduct causing risks to the fetus. After all, if the husband’s interest in the fetus’ safety is a sufficient predicate for state regulation, the State could reasonably conclude that pregnant wives should notify their husbands before drinking alcohol or smoking. Perhaps married women should notify their husbands before using contraceptives or before undergoing any type of surgery that may have complications affecting the husband’s interest in his wife’s reproductive organs. And if a husband’s interest justifies notice in any of these cases, one might reasonably argue that it justifies exactly what the Danforth Court held it did not justify — a requirement of the husband’s consent as well. A State may not give to a man the kind of dominion over his wife that parents exercise over their children.

Section 3209 embodies a view of marriage consonant with the common-law status of married women but repugnant to our present understanding of marriage and of the nature of the rights secured by the Constitution. Women do not lose their constitutionally protected liberty when they marry. The Constitution protects all individuals, male or female, married or unmarried, from the abuse of governmental power, even where that power is employed for the supposed benefit of a member of the individual’s family. These considerations confirm our conclusion that § 3209 is invalid.

*899D

We next consider the parental consent provision. Except in a medical emergency, an unemancipated young woman under 18 may not obtain an abortion unless she and one of her parents (or guardian) provides informed consent as defined above. If neither a parent nor a guardian provides consent, a court may authorize the performance of an abortion upon a determination that the young woman is mature and capable of giving informed consent and has in fact given her informed consent, or that an abortion would be in her best interests.

We have been over most of this ground before. Our cases establish, and we reaffirm today, that a State may require a minor seeking an abortion to obtain the consent of a parent or guardian, provided that there is an adequate judicial bypass procedure. See, e. g., Akron II, 497 U. S., at 510-519; Hodgson, 497 U. S., at 461 (O’Connor, J., concurring in part and concurring in judgment in part); id., at 497-501 (Kennedy, J., concurring in judgment in part and dissenting in part); Akron I, 462 U. S., at 440; Bellotti II, 443 U. S., at 643-644 (plurality opinion). Under these precedents, in our view, the one-parent consent requirement and judicial bypass procedure are constitutional.

The only argument made by petitioners respecting this provision and to which our prior decisions do not speak is the contention that the parental consent requirement is invalid because it requires informed parental consent. For the most part, petitioners’ argument is a reprise of their argument with respect to the informed consent requirement in general, and we reject it for the reasons given above. Indeed, some of the provisions regarding informed consent have particular force with respect to minors: the waiting period, for example, may provide the parent or parents of a pregnant young woman the opportunity to consult with her in private, and to discuss the consequences of her decision in *900the context of the values and moral or religious principles of their family. See Hodgson, supra, at 448-449 (opinion of Stevens, J.).

E

Under the recordkeeping and reporting requirements of the statute, every facility which performs abortions is required to file a report stating its name and address as well as the name and address of any related entity, such as a controlling or subsidiary organization. In the case of state-funded institutions, the information becomes public.

For each abortion performed, a report must be filed identifying: the physician (and the second physician where required); the facility; the referring physician or agency; the woman’s age; the number of prior pregnancies and prior abortions she has had;'gestational age; the type of abortion procedure; the date of the abortion; whether there were any pre-existing medical conditions which would complicate pregnancy; medical complications with the abortion; where applicable, the basis for the determination that the abortion was medically necessary; the weight of the aborted fetus; and whether the woman was married, and if so, whether notice was provided or the basis for the failure to give notice. Every abortion facility must also file quarterly reports showing the number of abortions performed broken down by trimester. See 18 Pa. Cons. Stat, §§ 3207, 8214 (1990). In all events, the identity of each woman who has had an abortion remains confidential.

In Danforth, 428 U. S., at 80, we held that recordkeeping and reporting provisions “that are reasonably directed to the preservation of maternal health and that properly respect a patient’s confidentiality and privacy are permissible.” We think that under this standard, all the provisions at issue here, except that relating to spousal notice, are constitutional. Although they do not relate to the State’s interest in informing the woman’s choice, they do relate to health. The collection of information with respect to actual patients *901is a vital element of medical research, and so it cannot be said that the requirements serve no purpose other than to make abortions more difficult. Nor do we find that the requirements impose a substantial obstacle to a woman’s choice. At most they might increase the cost of some abortions by a slight amount. While at some point increased cost could become a substantial obstacle, there is no such showing on the record before us.

Subsection (12) of the reporting provision requires the re-, porting of, among other things, a married woman’s “reason for failure to provide notice” to her husband. § 3214(a)(12). This provision in effeet requires women, as a condition of obtaining an abortion, to provide the Commonwealth with the precise information we have already recognized that many women have pressing reasons not to reveal. Like the spousal notice requirement itself, this provision places an undue burden on a woman’s choice, and must be invalidated for that reason.

VI

Our Constitution is a covenant running from the first generation of Americans to us and then to future generations. It is a coherent succession. Each generation must learn anew that the Constitution’s written terms embody ideas and aspirations that must survive more ages than one. We accept our responsibility not to retreat from interpreting the full meaning of the covenant in light of all of our precedents. We invoke it once again to define the freedom guaranteed by the Constitution’s own promise, the promise of liberty.

* * *

The judgment in No. 91-902 is affirmed. The judgment in No. 91-744 is affirmed in part and reversed in part, and the case is remanded for proceedings consistent with this opinion, including consideration of the question of severability.

It is so ordered.

*902APPENDIX TO OPINION OF O’CONNOR, KENNEDY, AND SOUTER, JJ.

Selected Provisions of the 1988 and 1989 Amendments to the Pennsylvania Abortion Control Act of 1982

18 PA. CONS. STAT. (1990).

Ҥ3203. Definitions.

“ ‘Medical emergency.’ That condition which, on the basis of the physician’s good faith clinical judgment, so complicates the medical condition of a pregnant woman as to necessitate the immediate abortion of her pregnancy to avert her death or for which a delay will create serious risk of substantial and irreversible impairment of major bodily function.”

Ҥ3205. Informed consent.

“(a) General rule. — No abortion shall be performed or induced except with the voluntary and informed consent of the woman upon whom the abortion is to be performed or induced. Except in the case of a medical emergency, consent to an abortion is voluntary and informed if and only if:

“(1) At least 24 hours prior to the abortion, the physician who is to perform the abortion or the referring physician has orally informed the woman of:

“(i) The nature of the proposed procedure or treatment and of those risks and alternatives to the procedure or treatment that a reasonable patient would consider material to the decision of whether or not to undergo the abortion.

“(ii) The probable gestational age of the unborn child at the time the abortion is to be performed.

“(iii) The medical risks associated with carrying her child to term.

“(2) At least 24 hours prior to the abortion, the physician who is to perform the abortion or the referring physician, or a qualified physician assistant, health care practitioner, technician or social worker to whom the re*903sponsibility has been delegated by either physician, has informed the pregnant woman that:

“(i) The department publishes printed materials which describe the unborn child and list agencies which offer alternatives to abortion and that she has a right to review the printed materials and that a copy will be provided to her free of charge if she chooses to review it.

“(ii) Medical assistance benefits may be available for prenatal care, childbirth and neonatal care, and that more detailed information on the availability of such assistance is contained in the printed materials published by the department.

“(iii) The father of the unborn child is liable to assist in the support of her child, even in instances where he has offered to pay for the abortion. In the case of rape, this information may be omitted.

“(3) A copy of the printed materials has been provided to the woman if she chooses to view these materials.

“(4) The pregnant woman certifies in writing, prior to the abortion, that the information required to be provided under paragraphs (1), (2) and (3) has been provided.

“(b) Emergency. — Where a medical emergency compels the performance of an abortion, the physician shall inform the woman, prior to the abortion if possible, of the medical indications supporting his judgment that an abortion is necessary to avert her death or to avert substantial and irreversible impairment of major bodily function.

“(e) Penalty. — Any physician who violates the provisions of this section is guilty of ‘unprofessional conduct’ and his license for the practice of medicine and surgery shall be subject to suspension or revocation in accordance with procedures provided under the act of October 5,1978 (P. L. 1109, No. 261), known as the Osteopathic Medical Practice Act, the *904act of December 20, 1985 (P. L. 457, No. 112), known as the Medical Practice Act of 1985, or their successor acts. Any physician who performs or induces an abortion without first obtaining the certification required by subsection (a)(4) or with knowledge or reason to know that the informed consent of the woman has not been obtained shall for the first offense be guilty of a summary offense and for each subsequent offense be guilty of a misdemeanor of the third degree. No physician shall be guilty of violating this section for failure to furnish the information required by subsection (a) if he or she can demonstrate, by a preponderance of the evidence, that he or she reasonably believed that furnishing the information would have resulted in a severely adverse effect on the physical or mental health of the patient.

“(d) Limitation on civil liability. — Any physician who complies with the provisions of this section may not be held civilly liable to his patient for failure to obtain informed consent to the abortion within the meaning of that term as defined by the act of October 15, 1975 (P. L. 390, No. Ill), known as the Health Care Services Malpractice Act.”

Ҥ 3206. Parental consent.

“(a) General rule. — Except in the case of a medical emergency or except as provided in this section, if a pregnant woman is less than 18 years of age and not emancipated, or if she has been adjudged an incompetent under 20 Pa. C. S. §5511 (relating to petition and hearing; examination by court-appointed physician), a physician shall not perform an abortion upon her unless, in the case of a woman who is less than 18 years of age, he first obtains the informed consent both of the pregnant woman and of one of her parents; or, in the case of a woman who is incompetent, he first obtains the informed consent of her guardian. In deciding whether to grant such consent, a pregnant woman’s parent or guardian shall consider only their child’s or ward’s best interests. In the case of a pregnancy that is the result of incest, where *905the father is a party to the incestuous act, the pregnant woman need only obtain the consent of her mother.

“(b) Unavailability of parent or guardian. — If both parents have died or are otherwise unavailable to the physician within a reasonable time and in a reasonable manner, consent of the pregnant woman’s guardian or guardians shall be sufficient. If the pregnant woman’s parents are divorced, consent of the parent having custody shall be sufficient. If neither any parent nor a legal guardian is available to the physician within a reasonable time and in a reasonable manner, consent of any adult person standing in loco parentis shall be sufficient.

“(c) Petition to the court for consent. — If both of the parents or guardians of the pregnant woman refuse to consent to the performance of an abortion or if she elects not to seek the consent of either of her parents or of her guardian, the court of common pleas of the judicial district in which the applicant resides or in which the abortion is sought shall, upon petition or motion, after an appropriate hearing, authorize a physician to perform the abortion if the court determines that the pregnant woman is mature and capable of giving informed consent to the proposed abortion, and has, in fact, given such consent.

“(d) Court order. — If the court determines that the pregnant woman is not mature and capable of giving informed consent or if the pregnant woman does not claim to be mature and capable of giving informed consent, the court shall determine whether the performance of an abortion upon her would be in her best interests. If the court determines that the performance of an abortion would be in the best interests of the woman, it shall authorize a physician to perform the abortion.

“(e) Representation in proceedings. — The pregnant woman may participate in proceedings in the court on her own behalf and the court may appoint a guardian ad litem to assist her. The court shall, however, advise her that she has *906a right to court appointed counsel, and shall provide her with such counsel unless she wishes to appear with private counsel or has knowingly and intelligently waived representation by counsel.”

Ҥ3207. Abortion facilities.

“(b) Reports. — Within 30 days after the effective date of this chapter, every facility at which abortions are performed shall file, and update immediately upon any change, a report with the department, containing the following information:

“(1) Name and address of the facility.

“(2) Name and address of any parent, subsidiary or affiliated organizations, corporations or associations.

“(3) Name and address of any parent, subsidiary or affiliated organizations, corporations or associations having contemporaneous commonality of ownership, beneficial interest, directorship or offieership with any other facility.

The information contained in those reports which are filed pursuant to this subsection by facilities which receive State-appropriated funds during the 12-ealendar-month period immediately preceding a request to inspect or copy such reports shall be deemed public information. Reports filed by facilities which do not receive State-appropriated funds shall only be available to law enforcement officials, the State Board of Medicine and the State Board of Osteopathic Medicine for use in the performance of their official duties. Any facility failing to comply with the provisions of this subsection shall be assessed by the department a fine of $500 for each day it is in violation hereof.”

Ҥ 3208. Printed information.

“(a) General rule. — The department shall cause to be published in English, Spanish and Vietnamese, within 60 days after this chapter becomes law, and shall update on an annual basis, the following easily comprehensible printed materials:

*907“(1) Geographically indexed materials designed to inform the woman of public and private agencies and services available to assist a woman through pregnancy, upon childbirth and while the child is dependent, including adoption agencies, which shall include a comprehensive list of the agencies available, a description of the services they offer and a description of the manner, including telephone numbers, in which they might be contacted, or, at the option of the department, printed materials including a toll-free 24-hour a day telephone number which may be called to obtain, orally, such a list and description of agencies in the locality of the caller' and of the services they offer. The materials shall provide information on the availability of medical assistance benefits for prenatal care, childbirth and neonatal care, and state that it is unlawful for any individual to coerce a woman to undergo abortion, that any physician who performs an abortion upon a woman without obtaining her informed consent or without according her a private medical consultation may be liable to her for damages in a civil action at law, that the father of a child is liable to assist in the support of that child, even in instances where the father has offered to pay for an abortion and that the law permits adoptive parents to pay costs of prenatal care, childbirth and neonatal care.

“(2) Materials designed to inform the woman of the probable anatomical and physiological characteristics of the unborn child at two-week gestational increments from fertilization to full term, including pictures representing the development of unborn children at two-week gestational increments, and any relevant information on the possibility of the unborn child’s survival; provided that any such pictures or drawings must contain the dimensions of the fetus and must be realistic and appropriate for the woman’s stage of pregnancy. The materials shall be objective, non-judgmental and designed *908to convey only accurate scientific information about the unborn child at the various gestational ages.' The material shall also contain objective information describing the methods of abortion procedures commonly employed, the medical risks commonly associated with each such procedure, the possible detrimental psychological effects of abortion and the medical risks commonly associated with each such procedure and the medical risks commonly associated with carrying a child to term.

“(b) Format. — The materials shall be printed in a typeface large enough to be clearly legible.

“(e) Free distribution. — The materials required under this section shall be available at no cost from the department upon request and in appropriate number to any person, facility or hospital.”

Ҥ3209. Spousal notice.

“(a) Spousal notice required. — In order to further the Commonwealth’s interest in promoting the integrity of the marital relationship and to protect a spouse’s interests in having children within marriage and in protecting the prenatal life of that spouse’s child, no physician shall perform an abortion on a married woman, except as provided in subsections (b) and (c), unless he or she has received a signed statement, which need not be notarized, from the woman upon whom the abortion is to be performed, that she has notified her spouse that she is about to undergo an abortion. The statement shall bear a notice that any false statement made therein is punishable by law.

“(b) Exceptions. — The statement certifying that the notice required by subsection (a) has been given need not be furnished where the woman provides the physician a signed statement certifying at least one of the following:

“(1) Her spouse is not the father of the child.

“(2) Her spouse, after diligent effort, could not be located.

*909“(3) The pregnancy is a result of spousal sexual assault as described in section 3128 (relating to spousal sexual assault), which has been reported to a law enforcement agency having the requisite jurisdiction.

“(4) The woman has reason to believe that the furnishing of notice to her spouse is likely to result in the infliction of bodily injury upon her by her spouse or by another individual.

Such statement need not be notarized, but shall bear a notice that any false statements made therein are punishable by law.

“(c) Medical emergency. — The requirements of subsection (a) shall not apply in case of a medical emergency.

“(d) Forms. — The department shall cause to be published, forms which may be utilized for purposes of providing the signed statements required by subsections (a) and (b). The department shall distribute an adequate supply of such forms to all abortion facilities in this Commonwealth.

“(e) Penalty; civil action. — Any'physician who violates the provisions of this section is guilty of 'unprofessional conduct,’ and his -or her license for the practice of medicine and surgery shall be subject to suspension or revocation in accordance with procedures provided under the act of October 5, 1978 (P. L. 1109, No. 261), known as the Osteopathic Medical Practice Act, the act of December 20, 1985 (P. L. 457, No. 112), known as the Medical Practice Act of 1985, or their successor acts. In addition, any physician who knowingly violates the provisions of this section shall be civilly liable to the spouse who is the father of the aborted child for any damages caused thereby and for punitive damages in the amount of $5,000, and the court shall award a prevailing plaintiff a reasonable attorney fee as part of costs.”

Ҥ3214. Reporting.

“(a) General rule. — For the purpose of promotion of maternal health and life by adding to the sum of medical and *910public health knowledge through the compilation of relevant data, and to promote the Commonwealth’s interest in protection of the unborn child, a report of each abortion performed shall be made to the department on forms prescribed by it. The report forms shall not identify the individual patient by name and shall include the following information:

“(1) Identification of the physician who performed the abortion, the concurring physician as required by section 3211(c)(2) (relating to abortion on unborn child of 24 or more weeks gestational age), the second physician as required by section 3211(c)(5) and the facility where the abortion was performed and of the referring physician, agency or service, if any.

“(2) The county and state in which the woman resides.

“(3) The woman’s age.

“(4) The number of prior pregnancies and prior abortions of the woman.

“(5) The gestational age of the unborn child at the time of the abortion.

“(6) The type of procedure performed or prescribed and the date of the abortion.

“(7) Pre-existing medical conditions of the woman which would complicate pregnancy, if any, and if known, any medical complication which resulted from the abortion itself.

“(8) The basis for the medical judgment of the physician who performed the abortion that the abortion was necessary to prevent either the death of the pregnant woman or the substantial and irreversible impairment of a major bodily function of the woman, where an abortion has been performed pursuant to section 3211(b)(1).

“(9) The weight of the aborted child for any abortion performed pursuant to section 3211(b)(1).

“(10) Basis for any medical judgment that a medical emergency existed which excused the physician from compliance with any provision of this chapter.

*911“(11) The information required to be reported under section 3210(a) (relating to determination of gestational age).

“(12) Whether the abortion was performed upon a married woman and, if so, whether notice to her spouse was given. If no notice to her spouse was given, the report shall also indicate the reason for failure to provide notice.

“(f) Report by facility. — Every facility in which an abortion is performed within this Commonwealth during any quarter year shall file with the department a report showing the total number of abortions performed within the hospital or other facility during that quarter year. This report shall also show the total abortions performed in each trimester of pregnancy. Any report shall be available for public inspection and copying only if the facility receives State-appropriated funds within the 12-calendar-month period immediately preceding the filing of the report. These reports shall be submitted on a form prescribed by the department which will enable, a facility to indicate whether or not it is receiving State-appropriated funds. If the facility indicates on the form that it is not receiving State-appropriated funds, the department shall regard its report as confidential unless it receives other evidence which causes it to conclude that the facility receives State-appropriated funds.”

Justice Stevens,

concurring in part and dissenting in part.

The portions of the Court’s opinion that I have joined are more important than those with which I disagree. I shall therefore first comment on significant areas of agreement, and then explain the limited character of my disagreement.

*912f-H

The Court is unquestionably correct in concluding that the doctrine of stare decisis has controlling significance in a case of this kind, notwithstanding an individual Justice’s concerns about the merits.1 The central holding of Roe v. Wade, 410 U. S. 113 (1973), has been a “part of our law” for almost two decades. Planned Parenthood of Central Mo. v. Danforth, 428 U. S. 52, 101 (1976) (Stevens, J., concurring in part and dissenting in part). It was a natural sequel to the protection of individual liberty established in Griswold v. Connecticut, 381 U. S. 479 (1965). See also Carey v. Population Services International, 431 U. S. 678, 687, 702 (1977) (White, J., concurring in part and concurring in result). The societal costs of overruling Roe at this late date would be enormous. Roe is an integral part of a correct understanding of both the concept of liberty and the basic equality of men and women.

Stare decisis also provides a sufficient basis for my agreement with the joint opinion’s reaffirmation of Roe’s post-viability analysis. Specifically, I accept the proposition that “[i]f the State is interested in protecting fetal life after viability, it may go so far as to proscribe abortion during that period, except when it is necessary to preserve the life or health of the mother.” 410 U. S., at 163-164; see ante, at 879.

I also accept what is implicit in the Court’s analysis, namely, a reaffirmation of Roe’s explanation oí-why the State’s obligation to protect the life or health of the mother *913must take precedence over any duty to the unborn. The Court in Roe carefully considered, and rejected, the State’s argument “that the fetus is a ‘person’ within the language and meaning of the Fourteenth Amendment.” 410 U. S., at 156. After analyzing the usage of “person” in the Constitution, the Court concluded that that word “has application only postnatally.” Id., at 157. Commenting on the contingent property interests of the unborn that are generally represented by guardians ad litem, the Court noted: “Perfection of the interests involved, again, has generally been contingent upon live birth. In short, the unborn have never been recognized in the law as persons in the whole sense.” Id., at 162. Accordingly, an abortion is not “the termination of life entitled to Fourteenth Amendment protection.” Id., at 159. From this holding, there was no dissent, see id., at 173; indeed, no Member of the Court has ever questioned this fundamental proposition. Thus, as a matter of federal constitutional law, a developing organism that is not yet a “person” does not have what is sometimes described as a “right to life.”2 This has been and, by the Court’s holding today, *914remains a fundamental premise of our constitutional law governing reproductive autonomy.

II

My disagreement with the joint opinion begins with its understanding of the trimester framework established in Roe. Contrary to the suggestion of the joint opinion, ante, at 876, it is not a “contradiction” to recognize that the State may have a legitimate interest in potential human life and, at the same time, to conclude that that interest does not justify the regulation of abortion before viability (although other interests, such as maternal health, may). The fact that the State’s interest is legitimate does not tell us when, if ever, that interest outweighs the pregnant woman’s interest in personal liberty. It is appropriate, therefore, to consider more carefully the nature of the interests at stake.

First, it is clear that, in order to be legitimate, the State’s interest must be secular; consistent with the First Amendment the State may not promote a theological or sectarian interest. See Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S. 747, 778 (1986) (Stevens, J., concurring); see generally Webster v. Reproductive Health Services, 492 U. S. 490, 563-572 (1989) (Stevens, J., concurring in part and dissenting in part). Moreover, as discussed above, the state interest in potential human life is not an interest in loco parentis, for the fetus is not a person.

Identifying the State’s interests — which the States rarely articulate with any precision — makes clear that the interest in protecting potential life is not grounded in the Constitution. It is, instead, an indirect interest supported by both humanitarian and pragmatic concerns. Many of our citizens believe that any abortion reflects an unacceptable disrespect for potential human life and that the performance of more *915than a million abortions each year is intolerable; many find third-trimester abortions performed when the fetus is approaching personhood particularly offensive. The State has a legitimate interest in minimizing such offense. The State may also have a broader interest in expanding the population,3 believing society would benefit from the services of additional productive citizens — or that the potential human lives might include the occasional Mozart or Curie. These are the kinds of concerns that comprise the State’s interest in potential human life.

In counterpoise is the woman’s constitutional interest in liberty. One aspect of this liberty is a right to bodily integrity, a right to control one’s person. See, e. g., Rochin v. California, 342 U. S. 165 (1952); Skinner v. Oklahoma ex rel. Williamson, 316 U. S. 535 (1942). This right is neutral on the question of abortion: The Constitution would be equally offended by an absolute requirement that all women undergo abortions as by an absolute prohibition on abortions. “Our whole constitutional heritage rebels at the thought of giving government the power to control men’s minds.” Stanley v. Georgia, 394 U. S. 557, 565 (1969). The same holds true for the power to control women’s bodies.

The woman’s constitutional liberty interest also involves her freedom to decide matters of the highest privacy and the most personal nature. Cf. Whalen v. Roe, 429 U. S. 589, *916598-600 (1977). A woman considering abortion faces “a difficult choice having serious and personal consequences of major importance to her own future — perhaps to the salvation of her own immortal soul.” Thornburgh, 476 U. S., at 781. The authority to make such traumatic and yet empowering decisions is an element of basic human dignity. As the joint opinion so eloquently demonstrates, a woman’s decision to terminate her pregnancy is nothing less than a matter of conscience.

Weighing the State’s interest in potential life and the woman’s liberty interest, I agree with the joint opinion that the State may “ ‘ “expres[s] a preference for normal childbirth,”’” that the State may take steps to ensure that a woman’s choice “is thoughtful and informed,” and that “States are free to enact laws to provide a reasonable framework for a woman to make a decision that has such profound and lasting meaning.” Ante, at 872-873. Serious questions arise, however, when a State attempts to “persuade the woman to choose childbirth over abortion.” Ante, at 878. Decisional autonomy must limit the State’s power to inject into a woman’s most personal deliberations its own views of what is best. The State may promote its preferences by funding childbirth, by creating and maintaining alternatives to abortion, and by espousing the virtues of family; but it must respect the individual’s freedom to make such judgments.

This theme runs throughout our decisions concerning reproductive freedom. In general, Roe’s requirement that restrictions on abortions before viability be justified by the State’s interest in maternal health has prevented States from interjecting regulations designed to influence a woman’s decision. Thus, we have upheld regulations of abortion that are not efforts to sway or direct a woman’s choice, but rather are efforts to enhance the deliberative quality of that decision or are neutral regulations on the health aspects of her decision. We have, for example, upheld regulations re*917quiring written informed consent, see Planned Parenthood of Central Mo. v. Danforth, 428 U. S. 52 (1976); limited rec-ordkeeping and reporting, see ibid.; and pathology reports, see Planned Parenthood Assn. of Kansas City, Mo., Inc. v. Ashcroft, 462 U. S. 476 (1983); as well as various licensing and qualification provisions, see, e. g., Roe, 410 U. S., at 150; Simopoulos v. Virginia, 462 U. S. 506 (1983). Conversely, we have consistently rejected state efforts to prejudice a woman’s choice, either by limiting the information available to her, see Bigelow v. Virginia, 421 U. S. 809 (1975), or by “requiring] the delivery of information designed ‘to influence the woman’s informed choice between abortion or childbirth.’ ” Thornburgh, 476 U. S., at 760; see also Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416, 442-449 (1983).

In my opinion, the principles established in this long line of eases and the wisdom reflected in Justice Powell’s opinion for the Court in Akron (and followed by the Court just six years ago in Thornburgh) should govern our decision today. Under these principles, Pa. Cons. Stat. § § 3205(a)(2)(i) — (iii) (1990) of the Pennsylvania statute are unconstitutional. Those sections require a physician or counselor to provide the woman with a range of materials clearly designed to persuade her to choose not to undergo the abortion. While the Commonwealth is free, pursuant to § 3208 of the Pennsylvania law, to produce and disseminate such material, the Commonwealth may not inject such information into the woman’s deliberations just as she is weighing such an important ehoice.

Under this same analysis, §§ 3205(a)(l)(i) and (iii) of the Pennsylvania statute are constitutional. Those sections, which require the physician to inform a woman of the nature and risks of the abortion procedure and the medical risks of carrying to term, are neutral requirements comparable to those imposed in other medical procedures. Those sections indicate no effort by the Commonwealth to influence the *918woman’s choice in any way. If anything, such requirements enhance, rather than skew, the woman’s decisionmaking.

I — I HH

The 24-hour waiting period required by §§ 3205(a)(l)-(2) of the Pennsylvania statute raises even more serious concerns. Such a requirement arguably furthers the Commonwealth’s interests in two ways, neither of which is constitutionally permissible.

First, it may be argued that the 24-hour delay is justified by the mere fact that it is likely to reduce the number of abortions, thus furthering the Commonwealth’s interest in potential life. But such an argument would justify any form of coercion that placed an obstacle in the woman’s path. The Commonwealth cannot further its interests by simply wearing down the ability of the pregnant woman to exercise her constitutional right.

Second, it can more reasonably be argued that the 24-hour delay furthers the Commonwealth’s interest in ensuring that the woman’s decision is informed and thoughtful. But there is no evidence that the mandated delay benefits women or that it is necessary to enable the physician to convey any relevant information to the patient. The mandatory delay thus appears to rest on outmoded and unacceptable assumptions about the decisionmaking capacity of women. While there are well-established and consistently maintained reasons for the Commonwealth to view with skepticism the ability of minors to make decisions, see Hodgson v. Minnesota, 497 U. S. 417, 449 (1990),4 none of those reasons applies to an *919adult woman’s decisionmaking ability. Just as we have left behind the belief that a woman must consult her husband before undertaking serious matters, see ante, at 895-898, so we must reject the notion that a woman is less capable of deciding matters of gravity. Cf. Reed v. Reed, 404 U. S. 71 (1971).

In the alternative, the delay requirement may be premised on the belief that the decision to terminate a pregnancy is presumptively wrong. This premise is illegitimate. Those who disagree vehemently about the legality and morality of abortion agree about one thing: The decision to terminate a pregnancy is profound and difficult. No person undertakes such a decision lightly — and States may not presume that a woman has failed to reflect adequately merely because her conclusion differs from the State’s preference. A woman who has, in the privacy of her thoughts and conscience, weighed the options and made her decision cannot be forced to reconsider all, simply because the State believes she has come to the wrong conclusion.5

*920Part of the constitutional liberty to choose is the equal dignity to which each of us is entitled. A woman who decides to terminate her pregnancy is entitled to the same respect as a woman who decides to carry the fetus to term. The mandatory waiting period denies women that equal respect.

IV

In my opinion, a correct application of the “undue burden” standard leads to the same conclusion concerning the constitutionality of these requirements. A state-imposed burden on the exercise of a constitutional right is measured both by its effects and by its character: A burden may be “undue” either because the burden is too severe or because it lacks a legitimate, rational justification.6

The 24-hour delay requirement fails both parts of this test. The findings of the District Court establish the severity of *921the burden that the 24-hour delay imposes on many pregnant women. Yet even in those cases in which the delay is not especially onerous, it is, in my opinion, “undue” because there is no evidence that such a delay serves a useful and legitimate purpose. As indicated above, there is no legitimate reason to require a woman who has agonized over her decision to leave the clinic or hospital and return again another day. While a general requirement that a physician notify her patients about the risks of a proposed medical procedure is appropriate, a rigid requirement that all patients wait 24 hours or (what is true in practice) much longer to evaluate the significance of information that is either common knowledge or irrelevant is an irrational and, therefore, “undue” burden.

The counseling provisions are similarly infirm. Whenever government commands private citizens to speak or to listen, careful review of the justification for that command is particularly appropriate. In these eases, the Pennsylvania statute directs that counselors provide women seeking abortions with information concerning alternatives to abortion, the availability of medical assistance benefits, and the possibility of child-support payments. §§ 3205(a)(2)(i) — (iii). The statute requires that this information be given to all women seeking abortions, including those for whom such information is clearly useless, such as those who are married, those who have undergone the procedure in the past and are fully aware of the options, and those who are fully convinced that abortion is their only reasonable option. Moreover, the statute requires physicians to inform all of their patients of “[t]he probable gestational age of the unborn child.” §3205(a)(l)(ii). This information is of little decisional value in most cases, because 90% of all abortions are performed during the first trimester7 when fetal age has less relevance than when the fetus nears viability. Nor can the informa*922tion required by the statute be justified as relevant to any “philosophic” or “social” argument, ante, at 872, either favoring or disfavoring the abortion decision in a particular ease. In light of all of these facts, I conclude that the information requirements in §3205(a)(l)(ii) and §§ 3205(a)(2)(i)-(iii) do not serve a useful purpose and thus constitute an unnecessary— and therefore undue — burden on the woman’s constitutional liberty to decide to terminate her pregnancy.

Accordingly, while I disagree with Parts IV, V-B, and V-D of the joint opinion,8 I join the remainder of the Court’s opinion.

Justice Blackmun,

concurring in part, concurring in the judgment in part, and dissenting in part.

I join Parts I, II, III, V-A, V-C, and VI of the joint opinion of Justices O’Connor, Kennedy, and Souter, ante.

Three years ago, in Webster v. Reproductive Health Services, 492 U. S. 490 (1989), four Members of this Court appeared poised to “cas[t] into darkness the hopes and visions of every woman in this country” who had come to believe that the Constitution guaranteed her the right to reproductive choice. Id., at 557 (Blackmun, J., dissenting). See id., at 499 (plurality opinion of Rehnquist, C. J., joined by White and Kennedy, JJ.); id., at 532 (Scalia, J., concurring in part and concurring in judgment). All that remained between the promise of Roe and the darkness of the plurality was a single, flickering flame. Decisions since Webster gave little reason to hope that this flame would cast much light. See, e. g., Ohio v. Akron Center for Reproductive Health, 497 U. S. 502, 524 (1990) (Blackmun, J., dissenting). But now, just when so many expected the darkness to fall, the flame has grown bright.

*923I do not underestimate the significance of today’s joint opinion. Yet I remain steadfast in my belief that the right to reproductive choice is entitled to the full protection afforded by this Court before Webster. And I fear for the darkness as four Justices anxiously await the single vote necessary to extinguish the light.

I — i

Make no mistake, the joint opinion of Justices O’Connor, Kennedy, and Souter is an act of personal courage and constitutional principle. In contrast to previous decisions in which Justices O’Connor and Kennedy postponed reconsideration of Roe v. Wade, 410 U. S. 113 (1973), the authors of the joint opinion today join Justice Stevens and me in concluding that “the essential holding of Roe v. Wade should be retained and once again reaffirmed.” Ante, at 846. In brief, five Members of this Court today recognize that “the Constitution protects a woman’s right to terminate her pregnancy in its early stages.” Ante, at 844.

A fervent view of individual liberty and the force of stare decisis have led the Court to this conclusion. Ante, at 853. Today a majority reaffirms that the Due Process Clause of the Fourteenth Amendment establishes “a realm of personal liberty which the government may not enter,” ante, at 847— a realm whose outer limits cannot be determined by interpretations of the Constitution that focus only on the specific practices of States at the time the Fourteenth Amendment was adopted. See ante, at 848-849. Included within this realm of liberty is “The right of the individual, married or single, to be free from unwarranted governmental intrusion into matters so fundamentally affecting a person as the decision whether to bear or beget a child.’ ” Ante, at 851, quoting Eisenstadt v. Baird, 405 U. S. 438, 453 (1972) (emphasis in original). “These matters, involving the most intimate and personal choices a person may make in a lifetime, choices central to personal dignity and autonomy, are central to the *924liberty protected by the Fourteenth Amendment.” Ante, at 851 (emphasis added). Finally, the Court today recognizes that in the case of abortion, “the liberty of the woman is at stake in a sense unique to the human condition and so unique to the law. The mother who carries a child to full term is subject to anxieties, to physical constraints, to pain that only she must bear.” Ante, at 852.

The Court’s reaffirmation of Roe’s central holding is also based on the force of stare decisis. “[N]o erosion of principle going to liberty or personal autonomy has left Roe’s central holding a doctrinal remnant; Roe portends no developments at odds with other precedent for the analysis of personal liberty; and no changes of fact have rendered viability more or less appropriate as the point at which the balance of interests tips.” Ante, at 860-861. Indeed, the Court acknowledges that Roe’s limitation on state power could not be removed “without serious inequity to those who have relied upon it or significant damage to the stability of the society governed by it.” Ante, at 855. In the 19 years since Roe was decided, that case has shaped more than reproductive planning — “[a]n entire generation has come of age free to assume Roe’s concept of liberty in defining the capacity of women to act in society, and to make reproductive decisions.” Ante, at 860. The Court understands that, having “call[ed] the contending sides ... to end their national division by accepting a common mandate rooted in the Constitution,” ante, at 867, a decision to overrule Roe “would seriously weaken the Court’s capacity to exercise the judicial power and to function as the Supreme Court of a Nation dedicated to the rule of law.” Ante, at 865. What has happened today should serve as a model for future Justices and a warning to all who have tried to turn this Court into yet another political branch.

In striking down the Pennsylvania statute’s spousal notification requirement, the Court has established a framework *925for evaluating abortion regulations that responds to the social context of women facing issues of reproductive choice.1 In determining the burden imposed by the challenged regulation, the Court inquires whether the regulation’s “purpose or effect is to plaee a substantial obstacle in the path of a woman seeking an abortion before the fetus attains viability.” Ante, at 878 (emphasis added). The Court reaffirms: “The proper focus of constitutional inquiry is the group for whom the law is a restriction, not the group for whom the law is irrelevant.” Ante, at 894. Looking at this group, the Court inquires, based on expert testimony, empirical studies, and common sense, whether “in a large fraction of the cases in which [the restriction] is relevant, it will operate as a substantial obstacle to a woman’s choice to undergo an abortion.” Ante, at 895. “A statute with this purpose is invalid because the means chosen by the State to further the interest in potential life must be calculated to inform the woman’s free choice, not hinder it.” Ante, at 877. And in applying its test, the Court remains sensitive to the unique role of women in the decisionmaking process. Whatever may have been the practice when the Fourteenth Amendment was adopted, the Court observes, “[w]omen do not lose their constitutionally protected liberty when they marry. The Constitution protects all individuals, male or female, married or unmarried, from the abuse of governmental power, even where that power is employed for the supposed benefit of a member of the individual’s, family.” Ante, at 898.2

*926Lastly, while I believe that the joint opinion errs in failing to invalidate the other regulations, I am pleased that the joint opinion has not ruled out the possibility that these regulations may be shown to impose an unconstitutional burden. The joint opinion makes clear that its specific holdings are based on the insufficiency of the record before it. See, e. g., ante, at 885-886. I am confident that in the future evidence will be produced to show that “in a large fraction of the cases in which [these regulations are] relevant, [they] will operate as a substantial obstacle to a woman’s choice to undergo an abortion.” Ante, at 895.

II

Today, no less than yesterday, the Constitution and decisions of this Court require that a State’s abortion restrictions be subjected to the strictest judicial scrutiny. Our precedents and the joint opinion’s principles require us to subject all non-de-minimis abortion regulations to strict scrutiny. Under this standard, the Pennsylvania statute’s provisions requiring content-based counseling, a 24-hour delay, informed parental consent, and reporting of abortion-related information must be invalidated.

A

The Court today reaffirms the long recognized rights of privacy and bodily integrity. As early as 1891, the Court held, “[n]o right is held more sacred, or is more carefully guarded by the common law, than the right of every individual to the possession and control of his own person, free from all restraint or interference of others ... .” Union Pacific R. Co. v. Botsford, 141 U. S. 250, 251 (1891). Throughout this century, this Court also has held that the fundamental right of privacy protects citizens against governmental in-*927trusión in such intimate family matters as procreation, child-rearing, marriage, and contraceptive choice. See ante, at 847-849. These eases embody the principle that personal decisions that profoundly affeet bodily integrity, identity, and destiny should be largely beyond the reach of government. Eisenstadt, 405 U. S., at 453. In Roe v. Wade, this Court correctly applied these principles to a woman’s right to choose abortion.

State restrictions on abortion violate a woman’s right of privacy in two ways. First, compelled continuation of a pregnancy infringes upon a woman’s right to bodily integrity by imposing substantial physical intrusions and significant risks of physical harm. During pregnancy, women experience dramatic physical changes and a wide range of health consequences. Labor and delivery pose additional health risks and physical demands. In short, restrictive abortion laws force women to endure physical invasions far more substantial than those this Court has held to violate the constitutional principle of bodily integrity in other contexts. See, e. g., Winston v. Lee, 470 U. S. 753 (1985) (invalidating surgical removal of bullet from murder suspect); Rochin v. California, 342 U. S. 165 (1952) (invalidating stomach pumping).3

Further, when the State restricts a woman’s right to terminate her pregnancy, it deprives a woman of the right to make her own decision about reproduction and family planning — critical life choices that this Court long has deemed central to the right to privacy. The decision to terminate or continue a pregnancy has no less an impact on a woman’s life than decisions about contraception or marriage. 410 U. S., *928at 153. Because motherhood has a dramatic impact on a woman’s educational prospects, employment opportunities, and self-determination, restrictive abortion laws deprive her of basic control over her life.- For these reasons, “the decision whether or not to beget or bear a child” lies at “the very heart of this cluster of constitutionally protected choices.” Carey v. Population Services International, 431 U. S. 678, 685 (1977).

A State’s restrictions on a woman’s right to terminate her pregnancy also implicate constitutional guarantees of gender equality. State restrictions on abortion compel women to continue pregnancies they otherwise might terminate. By restricting the right to terminate pregnancies, the State conscripts women’s bodies into its service, forcing women to continue their pregnancies, suffer the pains of childbirth, and in most instances, provide years of maternal care. The State does not compensate women for their services; instead, it assumes that they owe this duty as a matter of course. This assumption — that women can simply be forced to accept the “natural” status and incidents of motherhood — appears to rest upon a conception of women’s role that -has triggered the protection of the Equal Protection Clause. See, e.g., Mississippi Univ. for Women v. Hogan, 458 U. S. 718, 724-726 (1982); Craig v. Boren, 429 U. S. 190, 198-199 (1976).4 The joint opinion recognizes that these assumptions about women’s place in society “are no longer consistent with our *929understanding of the family, the individual, or the Constitution.” Ante, at 897.

B

The Court has held that limitations on the right of privacy are permissible only if they survive “strict” constitutional scrutiny — that is, only if the governmental entity imposing the restriction can demonstrate that the limitation is both necessary and narrowly tailored to serve a compelling governmental interest. Griswold v. Connecticut, 381 U. S. 479, 485 (1965). We have applied this principle specifically in the context of abortion regulations. Roe v. Wade, 410 U. S., at 155.5

Roe implemented these principles through a framework that was designed “to ensure that the woman’s right to choose not become so subordinate to the State’s interest in promoting fetal life that her choice exists in theory but not in fact,” ante, at 872. Roe identified two relevant state interests: “an interest in preserving and protecting the health of the pregnant woman” and an interest in “protecting the potentiality of human life.” 410 U. S., at 162. With respect to the State’s interest in the health of the mother, “the ‘compelling’ point ... is at approximately the end of the first trimester,” because it is at that point that the mortality rate in abortion approaches that in childbirth. Id., at 163. With respect to the State’s interest in potential life, “the ‘compelling’ point is at viability,” because it is at that point that the *930fetus “presumably has the capability of meaningful life outside the mother’s womb.” Ibid. In order to fulfill the requirement of narrow tailoring, “the State is obligated to make a reasonable effort to limit the effect of its regulations to the period in the trimester during which its health interest will be furthered.” Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416, 434 (1983).

In my view, application of this analytical framework is no less warranted than when it was approved by seven Members of this Court in Roe. Strict scrutiny of state limitations on reproductive choice still offers the most secure protection of the woman’s right to make her own reproductive decisions, free from state coercion. No majority of this Court has ever agreed upon an alternative approach. The factual premises of the trimester framework have not been undermined, see Webster, 492 U. S., at 553 (Blackmun, J., dissenting), and the Roe framework is far more administra-ble, and far less manipulate, than the “undue burden” standard adopted by the joint opinion.

Nonetheless, three criticisms of the trimester framework continue to be uttered. First, the trimester framework is attacked because its key elements do not appear in the text of the Constitution. My response to this attack remains the same as it was in Webster:

“Were this a true concern, we would have to abandon most of our constitutional jurisprudence. [Tjhe ‘critical elements’ of countless constitutional doctrines nowhere appear in the Constitution’s text.... The Constitution makes no mention, for example, of the First Amendment’s ‘actual malice’ standard for proving certain libels, see New York Times Co. v. Sullivan, 376 U. S. 254 (1964). ... Similarly, the Constitution makes no mention of the rational-basis test, or the specific verbal formulations of intermediate and strict scrutiny by which this Court evaluates claims under the Equal Protection Clause. The reason is simple. Like the Roe framework, these *931tests or standards are not, and do not purport to be, rights protected by the Constitution. Rather, they are judge-made methods for evaluating and measuring the strength and scope of constitutional rights or for balancing the constitutional rights of individuals against the competing interests of government.” Id., at 548.

The second criticism is that the framework more closely resembles a regulatory code than a body of constitutional doctrine. Again, my answer remains the same as in Webster:

“[I]f this were a true and genuine concern, we would have to abandon vast areas of our constitutional jurisprudence. . . . Are [the distinctions entailed in the trimester framework] any finer, or more ‘regulatory,’ than the distinctions we have often drawn in our First Amendment jurisprudence, where, for example, we have held that a ‘release time’ program permitting public-school students to leave school grounds during school hours to receive religious instruction does not violate the Establishment Clause, even though a release-time program permitting religious instruction on school grounds does violate the Clause? Compare Zorach v. Clauson, 343 U. S. 306 (1952), with Illinois ex rel. Mc-Collum v. Board of Education of School Dist. No. 71, Champaign County, 333 U. S. 203 (1948)... . Similarly, in a Sixth Amendment case, the Court held that although an overnight ban on attorney-client communication violated the constitutionally guaranteed right to counsel, Geders v. United States, 425 U. S. 80 (1976), that right was not violated when a trial judge separated a defendant from his lawyer during a 15-minute recess after the defendant’s direct testimony. Perry v. Leeke, 488 U. S. 272 (1989).

“That numerous constitutional doctrines result in narrow differentiations between similar circumstances does *932not mean that this Court has abandoned adjudication in favor of regulation.” Id., at 549-550.

The final, and more genuine, criticism of the trimester framework is that it fails to find the State’s interest in potential human life compelling throughout pregnancy. No Member of this Court — nor for that matter, the Solicitor General, see Tr. of Oral Arg. 42 — has ever questioned our holding in Roe that an abortion is not “the termination of life entitled to Fourteenth Amendment protection.” 410 U. S., at 159. Accordingly, a State’s interest in protecting fetal life is not grounded in the Constitution. Nor, consistent with our Establishment Clause, can it be a theological or sectarian interest. See Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S. 747, 778 (1986) (Stevens, J., concurring). It is, instead, a legitimate interest grounded in humanitarian or pragmatic concerns. See ante, at 914-915 (Stevens, J., concurring in part and dissenting in part).

But while a State has “legitimate interests from the outset of the pregnancy in protecting the health of the worn an and the life of the fetus that may become a child,” ante, at 846, legitimate interests are not enough. To overcome the burden of strict scrutiny, the interests must be compelling. The question then is how best to accommodate the State’s interest in potential human life with the constitutional liberties of pregnant women. Again, I stand by the views I expressed in Webster:

“I remain convinced, as six other Members of this Court 16 years ago were convinced, that the Roe framework, and the viability standard in particular, fairly, sensibly, and effectively functions to safeguard the constitutional liberties of pregnant women while recognizing and accommodating the State’s interest in potential human life. The viability line reflects the biological facts and truths of fetal development; it marks that threshold moment prior to which a fetus cannot survive separate from the *933woman and cannot reasonably and objectively be regarded as a subject of rights or interests distinct from, or paramount to, those of the pregnant woman. At the same time, the viability standard takes account of the undeniable fact that as the fetus evolves into its postnatal form, and as it loses its dependence on the uterine environment, the State’s interest in the fetus’ potential human life, and in fostering a regard for human life in general, becomes compelling. As a practical matter, because viability follows ‘quickening’ — the point at which. a woman feels movement in her womb — and because viability occurs no earlier than 23 weeks gestational age, it establishes an easily applicable standard for regulating abortion while providing a pregnant woman ample time to exercise her fundamental right with her responsible physician to terminate her pregnancy.” 492 U. S., at 553-554.6

Roe’s trimester framework does not ignore the State’s interest in prenatal life. Like Justice Stevens, ante, at 916,

I agree that the State may take steps to ensure that a woman’s choice “is thoughtful and informed,” ante, at 872, and that “States are free to enact laws to provide a reasonable framework for a woman to make a decision that has such profound and lasting meaning.” Ante, at 873. But

“[sjerious questions- arise ... when a State attempts to persuade the woman to choose childbirth over abortion. Ante, at 878. Decisional autonomy must limit the State’s power to inject into a woman’s most personal deliberations its own views of what is best. The State may promote its preferences by funding childbirth, by creating and maintaining alternatives to abortion, and by espousing the virtues of family; but it must respect *934the individual’s freedom to make such judgments.” Ante, at 916 (Stevens, J., concurring in part and dissenting in part) (internal quotation marks omitted).

As the joint opinion recognizes, “the means chosen by the State to further the interest in potential life must be calculated to inform the woman’s free choice, not hinder it.” Ante, at 877.

In sum, .Roe’s requirement of strict scrutiny as implemented through a trimester framework should not be disturbed. No other approach has gained a majority, and no other is more protective of the woman’s fundamental right. Lastly, no other approach properly accommodates the woman’s constitutional right with the State’s legitimate interests.

C

Application of the strict scrutiny standard results in the invalidation of all the challenged provisions. Indeed, as this Court has invalidated virtually identical provisions in prior cases, stare decisis requires that we again strike them down.

This Court has upheld informed- and written-consent requirements only where the State has demonstrated that they genuinely further important health-related state concerns. See Planned Parenthood of Central Mo. v. Danforth, 428 U. S. 52, 65-67 (1976). A State may not, under the guise of securing informed consent, “require the delivery of information ‘designed to influence the woman’s informed choice between abortion or childbirth.’ ” Thornburgh, 476 U. S., at 760, quoting Akron, 462 U. S., at 443-444. Rigid requirements that a specific body of information be imparted to a woman in all cases, regardless of the needs of the patient, improperly intrude upon the discretion of the pregnant woman’s physician and thereby impose an “‘undesired and uncomfortable straitjacket.’ ” Thornburgh, 476 U. S., at 762, quoting Danforth, 428 U. S., at 67, n. 8.

Measured against these principles, some aspects of the Pennsylvania informed-consent scheme are unconstitutional. *935While it is unobjectionable for the Commonwealth to require that the patient be informed of the nature of the procedure, the health risks of the abortion and of childbirth, and the probable gestational age of the unborn child, compare Pa. Cons. Stat. §§3205(a)(i)-(iii) (1990) with Akron, 462 U. S., at 446, n. 37, I remain unconvinced that there is a vital state need for insisting that the information be provided by a physician rather than a counselor. Id., at 448. The District Court found that the physician-only requirement necessarily would increase costs to the plaintiff clinics, costs that undoubtedly would be passed on to patients. And because trained women counselors are often more understanding than physicians, and generally havé more time to spend with patients, see App. 366-387, the physician-only disclosure requirement is not narrowly tailored to serve the Commonwealth’s interest in protecting maternal health.

Sections 3205(a)(2)(i)~(iii) of the Act further requires that the physician or a qualified nonphysieian inform the woman that printed materials are available from the Commonwealth that describe the fetus and provide information about medical assistance for childbirth, information about child support from the father, and a list of agencies offering adoption and other services as alternatives to abortion. Thornburgh invalidated biased patient-counseling requirements virtually identical to the one at issue here. What we said of those requirements fully applies in these cases:

“[T]he listing of agencies in the printed Pennsylvania form presents serious problems; it contains names of agencies that well may be out of step with the needs of the particular woman and thus places the physician in an awkward position and infringes upon his or her professional responsibilities. Forcing the physician or counselor to present the materials and the list to the woman makes him or her in effect an agent of the State in treating the woman and places his or her imprimatur upon both the materials and the list. All this is, or *936comes close to being, state medicine imposed upon the woman, not the professional medical guidance she seeks, and it officially structures — as it obviously was intended to do — the dialogue between the woman and her physician.

“The requirements . . . that the woman be advised that medical assistance benefits may be available, and that the father is responsible for financial assistance in the support of the child similarly are poorly disguised elements of discouragement for the abortion decision. Much of this ..., for many patients, would be irrelevant and inappropriate. For a patient with a life-threatening pregnancy, the ‘information’ in its very rendition may be cruel as well as destructive of the physician-patient relationship. As any experienced social worker or other counselor knows, theoretical financial responsibility often does not equate with fulfillment.... Under the guise of informed consent, the Act requires the dissemination of information that is not relevant to such consent, and, thus, it advances no legitimate state interest.” 476 U' S., at 762-763 (citation omitted).

“This type of compelled information is the antithesis of informed consent,” id., at 764, and goes far beyond merely describing the general subject matter relevant to the woman’s decision. “That the Commonwealth does not, and surely would not, compel similar disclosure of every possible peril of necessary surgery or of simple vaccination, reveals the anti-abortion character of the statute and its real purpose.” Ibid.7

*937The 24-hour waiting period following the provision of the foregoing information is also clearly unconstitutional. The District Court found that the mandatory 24-hour delay could lead to delays in excess of 24 hours, thus increasing health risks, and that it would require two visits to the abortion provider, thereby increasing travel time, exposure to further harassment, and financial cost. Finally, the District Court found that the requirement would pose especially significant burdens on women living in rural areas and those women that have difficulty explaining their whereabouts. 744 F. Supp. 1328,1378-1379 (ED Pa. 1990). In Akron this Court invalidated a similarly arbitrary or inflexible waiting period because, as here, it fiirthered no legitimate state interest.8

As Justice Stevens insightfully concludes, the mandatory delay rests either on outmoded or unacceptable assumptions about the decisionmaking capacity of women or the belief that the decision to terminate the pregnancy is *938presumptively wrong. Ante, at 918-919. The requirement that women consider this obvious and slanted information for an additional 24 hours contained in these provisions will only influence the womán’s decision in improper ways. The vast majority of women will know this information — of the few that do not, it is less likely that their minds will be changed by this information than it will be either by the realization that the State opposes their choice or the need once again to endure abuse and harassment on return to the clinic.9

Except in the case of a medical emergency, § 3206 requires a physician to obtain the informed consent of a parent or guardian before performing an abortion on an unemanci-pated minor or an incompetent woman. Based on evidence in the record, the District Court concluded that, in order to fulfill the informed-consent requirement, generally accepted medical principles would require an in-person visit by the parent to the facility. 744 F. Supp., at 1382. Although the Court “has recognized that the State has somewhat broader authority to regulate the activities of children than of adults,” the State nevertheless must demonstrate that there is a “significant state interest in conditioning an abortion... that is not present in the case of an adult.” Danforth, 428 U. S., at 74-75 (emphasis added). The requirement of an in-person visit would carry with it the risk of a delay of several days or possibly weeks, even where the parent is willing to consent. While the State has an interest in encouraging parental involvement-in the minor’s abortion decision, §3206 is not narrowly drawn to serve that interest.10

*939Finally, the Pennsylvania statute requires every facility performing abortions to report its activities to the Commonwealth. Pennsylvania contends that this requirement is valid under Danforth, in which this Court held that record-keeping and reporting requirements that are reasonably directed to the preservation of maternal health and that properly respect a patient’s confidentiality are permissible. Id., at 79-81. The Commonwealth attempts to justify its required reports on the ground that the public has a right to know how its tax dollars are spent. A regulation designed to inform the public about public expenditures does not further the Commonwealth’s interest in protecting maternal health. Accordingly, such a regulation cannot justify a legally significant burden on a woman’s right to obtain an abortion.

The confidential reports concerning the identities and medical judgment of physicians involved in abortions at first glance may seem valid, given the Commonwealth’s interest in maternal health and enforcement of the Act. The District Court found, however, that, notwithstanding the confidentiality protections, many physicians, particularly those who have previously discontinued performing abortions because of harassment, would refuse to refer patients to abortion clinics if their names were to appear on these reports. 744 F. Supp., at 1392. The Commonwealth has failed to show that the name of the referring physician either adds to the pool of scientific knowledge concerning abortion or is reasonably related to the Commonwealth’s interest in maternal health. I therefore agree with the District Court’s conclusion that the confidential reporting requirements are uneon-*940stitutional insofar as they require the name of the referring physician and the basis for his or her medical judgment.

In sum, I would affirm the judgment in No. 91-902 and reverse the judgment in No. 91-744 and remand the eases for further proceedings.

Ill

At long last, The ChieF Justice and those who have joined him admit it. Gone are the contentions that the issue need not be (or has not been) considered. There, on the first page, for all to see, is what was expected: “We believe that Roe was wrongly decided, and that it can and should be overruled consistently with our traditional approach to stare de-cisis in constitutional cases.” Post, at 944. If there is much reason to applaud the advances made by the joint opinion today, there is far more to fear from The ChieF Justice’s opinion.

The ChieF Justice’s criticism of Roe follows from his stunted conception of individual liberty. While recognizing that the Due Process Clause protects more than simple physical liberty, he then goes on to construe this Court’s personal-liberty cases as establishing only a laundry list of particular rights, rather than a principled account of how these particular rights are grounded in a more general right of privacy. Post, at 951. This constricted view is reinforced by The Chief Justice’s exclusive reliance on tradition as a source of fundamental rights. He argues that the record in favor of a right to abortion is no stronger than the record in Michael H. v. Gerald D., 491 U. S. 110 (1989), where the plurality found no fundamental right to visitation privileges by an adulterous father, or in Bowers v. Hardwick, 478 U. S. 186 (1986), where the Court found no fundamental right to engage in homosexual sodomy, or in a ease involving the “ ‘firing [of] a gun . . . into another person’s body.’” Post, at 951-952. In The Chief Justice’s world, a woman considering whether to terminate a pregnancy is entitled to no more protection than adulterers, murderers, and so-called sexual *941deviates.11 Given The ChieF Justice’s exclusive reliance on tradition, people using contraceptives seem the next likely candidate for his list of outcasts.

Even more shocking than The Chief Justice’s cramped notion of individual liberty is his complete omission of any discussion of the effects that compelled childbirth and motherhood have on women’s lives. The only expression of concern with women’s health is purely instrumental — for The CHIEF Justice, only women’s 'psychological health is a concern, and only to the extent that he assumes that every woman who decides to have an abortion does so without serious consideration of the moral implications of her decision. Post, at 967-968. In short, The Chief Justice’s view of the State’s compelling interest in maternal health has less to do with health than it does with compelling women to be maternal.

Nor does. The Chief Justice give any serious consideration to the doctrine of stare decisis. For The Chief Justice, the facts, that gave rise to Roe are surprisingly simple: “women become pregnant, there is a point somewhere, depending on medical technology, where a fetus becomes viable, and women give birth to children.” Post, at 965. This characterization of the issue thus allows The Chief Justice quickly to discard the joint opinion’s reliance argument by asserting that “reproductive planning could take virtually immediate account of” a decision overruling Roe. Post, at 956 (internal quotation marks omitted).

The Chief Justice’s narrow conception of individual liberty and stare decisis leads him to propose the same standard of review proposed by the plurality in Webster. “States may regulate abortion procedures in ways rationally related to a legitimate state interest. Williamson v. Lee Optical of Oklahoma, Inc., 348 U. S. 483, 491 (1955); cf. Stanley v. Illinois, 405 U.S. 645, 651-653 (1972).” Post, at 966. The *942CHIEF Justice then further weakens the test by providing an insurmountable requirement for facial challenges: Petitioners must “‘show that no set of circumstances exists under which the [provision] would be valid.’ ” Post, at 973, quoting Ohio v. Akron Center for Reproductive Health, 497 U. S., at 514. In short, in his view, petitioners must prove that the statute cannot constitutionally be applied to anyone. Finally, in applying his standard to the spousal-notification provision, The Chief Justice contends that the record lacks any “hard evidence” to support the joint opinion’s contention that a “large fraction” of women who prefer not to notify their husbands involve situations of battered women and unreported spousal assault. Post, at 974, n. 2. Yet throughout the explication of his standard, The Chief Justice never explains what hard evidence is, how large a fraction is required, or how a battered woman is supposed to pursue an as-applied challenge.

Under his standard, States can ban abortion if that ban is rationally related to a legitimate state interest — a standard which the United States calls “deferential, but not toothless.” Yet when' pressed at oral argument to describe the teeth, the best protection that the Solicitor General could offer to women was that a prohibition, enforced by criminal penalties, with no exception for the life of the mother, “could raise very serious questions.” Tr. of Oral Arg. 48. Perhaps, the Solicitor General offered, the failure to include an exemption for the life of the mother would be “arbitrary and capricious.” Id., at 49. If, as The Chief Justice contends, the undue burden test is made out of whole cloth, the so-called “arbitrary and capricious” limit is the Solicitor General’s “new clothes.”

Even if it is somehow “irrational” for a State to require a woman to risk her life for her child, what protection is offered for women who become pregnant through rape or incest? Is there anything arbitrary or capricious about a *943State’s prohibiting the sins of the father from being visited upon his offspring?12

But, we are reassured, there is always the protection of the democratic process. While there is much to be'praised about our democracy, our country since its founding has recognized that there are certain fundamental liberties that are not to be left to the whims of an election. A woman’s right to reproductive choice is one of those fundamental liberties. Accordingly, that liberty need not seek refuge at the ballot box.

IV

In one sense, the Court’s approach is worlds apart from that of The Chief Justice and Justíce Scalia. And yet, in another sense, the distance between the two approaches is short — the distance is but a single vote.

I am 83 years old. I cannot remain on this Court forever, and when I do step down, the confirmation process for my successor well may focus on the issue before us today. That, I regret, may be exactly where the choice between the two worlds will be made.

*944ChieF Justice Rehnquist,

with whom Justice White, Justice Scalia, and Justice Thomas join,

concurring in the judgment in part and dissenting in part.

The joint opinion, following its newly minted variation on stare decisis, retains the outer shell of Roe v. Wade, 410 U. S. 113 (1973), but beats a wholesale retreat from the substance of that case. We believe that Roe was wrongly decided, and that it can and should be overruled consistently with our traditional approach to stare decisis in constitutional cases. We would adopt the approach of the plurality in Webster v. Reproductive Health Services, 492 U. S. 490 (1989), and uphold the challenged provisions of the Pennsylvania statute in their entirety.

I

In ruling on this litigation below, the Court of Appeals for the Third Circuit first observed that “this appeal does not directly implicate Roe; this case involves the regulation of abortions rather than their outright prohibition.” 947 F. 2d 682, 687 (1991). Accordingly, the court directed its attention to the question of the standard of review for abortion regulations. In attempting to settle on the correct standard, however, the court confronted the confused state of this Court’s abortion jurisprudence. After considering the several opinions in Webster v. Reproductive Health Services, supra, and Hodgson v. Minnesota, 497 U. S. 417 (1990), the Court of Appeals concluded that Justice O’Connor’s “undue burden” test was controlling, as that was the narrowest ground on which we had upheld recent abortion regulations. 947 F. 2d, at 693-697 (“When a fragmented court decides a case and no single rationale explaining the result enjoys the assent of five Justices, the holding of the Court may be viewed as that position taken by those Members who concurred in the judgments on the narrowest grounds” (quoting Marks v. United States, 430 U. S. 188, 193 (1977) (internal quotation marks omitted))). Applying this standard, the Court of Appeals upheld all of the challenged regulations except the one *945requiring a woman to notify her spouse of an intended abortion.

In arguing that this Court should invalidate each of the provisions at issue, petitioners insist that we reaffirm our decision in Roe v. Wade, supra, in which we held unconstitutional a Texas statute making it a crime to procure an abortion except to save the life of the mother.1 We agree with the Court of Appeals that our decision in Roe is not directly implicated by the Pennsylvania statute, which does not prohibit, but simply regulates, abortion. But, as the Court of Appeals found, the state of our post-ftoe decisional law dealing with the regulation of abortion is confusing and uncertain, indicating that a reexamination of that line of eases is in order. Unfortunately for those who must apply this Court’s decisions, the reexamination undertaken today leaves the Court no less divided than beforehand. Although they reject the trimester framework that formed the underpinning of Roe, Justices O’Connor, Kennedy, and Souter adopt a revised undue burden standard to analyze the challenged regulations. We conclude, however, that such an outcome is an unjustified constitutional compromise, one which leaves the Court in a position to closely scrutinize all types of abortion regulations despite the fact that it lacks the power to do so under the Constitution.

In Roe, the Court opined that the State “does have an important and legitimate interest in preserving and protecting the health of the pregnant woman, . . . and that it has still another important and legitimate interest in protecting *946the potentiality of human life.” 410 U. S., at 162 (emphasis omitted). In the companion case of Doe v. Bolton, 410 U. S. 179 (1973), the Court referred to its conclusion in Roe “that a pregnant woman does not have an absolute constitutional right to an abortion on her demand.” 410 U. S., at 189. But while the language and holdings of these cases appeared to leave States free to regulate abortion procedures in a variety of ways, later decisions based on them have found considerably less latitude for such regulations than might have been expected.

For example, after Roe, many States have sought to protect their young citizens by requiring that a minor seeking an abortion involve her parents in the decision. Some States have simply required notification of the parents, while others have required a minor to obtain the consent of her parents. In a number of decisions, however, the Court has substantially limited the States in their ability to impose such requirements. With regard to parental notice requirements, we initially held that a State could require a minor to notify her parents before proceeding with an abortion. H. L. v. Matheson, 460 U. S. 398, 407-410 (1981). Recently, however, we indicated that a State’s ability to impose a notice requirement actually depends on whether it requires notice of one or both parents. We concluded that although the Constitution might allow a State to demand that notice be given to one parent prior to an abortion, it may not require that similar notice be given to two parents, unless the State incorporates a judicial bypass procedure in that two-parent requirement. Hodgson v. Minnesota, supra.

We have treated parental consent provisions even more harshly. Three years after Roe, we invalidated a Missouri regulation requiring that an unmarried woman under the age of 18 obtain the consent of one of her parents before proceeding with an abortion. We held that our abortion jurisprudence prohibited the State from imposing such a “blanket provision ... requiring the consent of a parent.” Planned Parenthood *947of Central Mo. v. Danforth, 428 U. S. 52, 74 (1976). In Bellotti v. Baird, 443 U. S. 622 (1979), the Court struck down a similar Massachusetts parental consent statute. A majority of the Court indicated, however, that a State could constitutionally require parental consent, if it alternatively allowed a pregnant minor to obtain an abortion without parental consent by showing either that she was mature enough to make her own decision, or that the abortion would be in her best interests. See id., at 643-644 (plurality opinion); id., at 656-657 (White, J., dissenting). In light of Bellotti, we have upheld one parental consent regulation which incorporated a judicial bypass option we viewed as sufficient, see Planned Parenthood Assn. of Kansas City, Mo., Inc. v. Ashcroft, 462 U. S. 476 (1983), but have invalidated another because of our belief that the judicial procedure did not satisfy the dictates of Bellotti, see Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416, 439-442 (1983). We have never had occasion, as we have in the parental notice context, to further parse our parental consent jurisprudence into one-parent and two-parent components.

In Roe, the Court observed that certain States recognized the right of the father to participate in the abortion decision in certain circumstances. Because neither Roe nor Doe involved the assertion of any paternal right, the Court expressly stated that the case did not disturb the validity of regulations that protected such a right. Roe v. Wade, supra, at 165, n. 67. But three years later, in Danforth, the Court extended its abortion jurisprudence and held that a State could not require that.a woman obtain the consent of her spouse before proceeding with an abortion. Planned Parenthood of Central Mo. v. Danforth, 428 U. S., at 69-71.

States have also regularly tried to ensure that a woman’s decision to have an abortion is an informed and well-considered one. In Danforth, we upheld a requirement that a woman sign a consent form prior to her abortion, and observed that “it is desirable and imperative that [the decision] *948be made with full knowledge of its nature and consequences.” Id., at 67. Since that case, however, we have twice invalidated state statutes designed to impart such knowledge to a woman seeking an abortion. In Akron, we held unconstitutional a regulation requiring a physician to inform a woman seeking an abortion of the status of her pregnancy, the development of her fetus, the date of possible viability, the complications that could result from an abortion, and the availability of agencies providing assistance and information with respect to adoption and childbirth. Akron v. Akron Center for Reproductive Health, supra, at 442-445. More recently, in Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S. 747 (1986), we struck down a more limited Pennsylvania regulation requiring that a woman be informed of the risks associated with the abortion procedure and the assistance available to her if she decided to proceed with her pregnancy, because we saw the compelled information as “the antithesis of informed consent.” Id., at 764. Even when a State has sought only to provide information that, in our view, was consistent with the Roe framework, we concluded that the State could not require that a physician furnish the information, but instead had to alternatively allow nonphysician counselors to provide it. Akron v. Akron Center for Reproductive Health, 462 U. S., at 448-449. In Akron as well, we went further and held that a State may not require a physician to wait 24 hours to perform an abortion after receiving the consent of a woman. Although the State sought to ensure that the' woman’s decision was carefully considered, the Court concluded that the Constitution forbade the State to impose any sort of delay. Id., at 449-451.

We have not allowed States much leeway to regulate even the actual abortion procedure. Although a State can require that second-trimester abortions be performed in outpatient clinics, see Simopoulos v. Virginia, 462 U. S. 506 (1983), we concluded in Akron and Ashcroft that a State could not *949require that such abortions be performed only in hospitals. See Akron v. Akron Center for Reproductive Health, supra, at 437-439; Planned Parenthood Assn. of Kansas City, Mo., Inc. v. Ashcroft, supra, at 481-482. Despite the fact that Roe expressly allowed regulation after the first trimester in furtherance of maternal health, “‘present medical knowledge,’” in our view, could not justify such a hospitalization requirement under the trimester framework. Akron v. Akron Center for Reproductive Health, supra, at 437 (quoting Roe v. Wade, supra, at 163). And in Danforth, the Court held that Missouri could not outlaw the saline amniocentesis method of abortion, concluding that the Missouri Legislature had “failed to appreciate and to consider several significant facts” in making its decision. 428 U. S., at 77.

Although Roe allowed state regulation after the point of viability to protect the potential life of the fetus, the Court subsequently rejected attempts to regulate in this manner. In Colautti v. Franklin, 439 U. S. 379 (1979), the Court struck down a statute that governed the determination of viability. Id., at 390-397. In the process, we made clear that the trimester framework incorporated only one definition of viability — ours—as we forbade States to decide that a certain objective indicator — “be it weeks of gestation or fetal weight or any other single factor” — should govern the definition of viability. Id., at 389. In that same case, we also invalidated a regulation requiring a physician to use the abortion technique offering the best chance for fetal survival when performing postviability abortions. See id., at 397-401; see also Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 768-769 (invalidating a similar regulation). In Thornburgh, the Court struck down Pennsylvania’s requirement that a second physician be present at postviability abortions to help preserve the health of the unborn child, on the ground that it did not incorporate a sufficient medical emergency exception. Id., at 769-771. Regulations governing the treatment of aborted fetuses have *950met a similar fate. In Akron, we invalidated a provision requiring physicians performing abortions to “insure that the remains of the unborn child are disposed of in a humane and sanitary manner.” 462 U. S., at 451 (internal quotation marks omitted).

Dissents in these cases expressed the view that the Court was expanding upon Roe in imposing ever greater restrictions on the States. See Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 783 (Burger, C. J., dissenting) (“The extent to which the Court has departed from the limitations expressed in Roe is readily apparent”); id., at 814 (White, J., dissenting) (“[T]he majority indiscriminately strikes down statutory provisions that in no way contravene the right recognized in Roe”). And, when confronted with state regulations of this type in past years, the Court has become increasingly more divided: The three most recent abortion cases have not commanded a Court opinion. See Ohio v. Akron Center for Reproductive Health, 497 U. S. 502 (1990); Hodgson v. Minnesota, 497 U. S. 417 (1990); Webster v. Reproductive Health Services, 492 U. S. 490 (1989).

The task of the Court of Appeals in the present eases was obviously complicated by this confusion and uncertainty. Following Marks v. United States, 480 U. S. 188 (1977), • it concluded that in light of Webster and Hodgson, the strict scrutiny standard enunciated in Roe was no longer applicable, and that the “undue burden” standard adopted by Justice O’Connor was the governing principle. This state of confusion and disagreement warrants reexamination of the “fundamental right” accorded to a woman’s decision to abort a fetus in Roe, with its concomitant requirement that any state regulation of abortion survive “strict scrutiny.” See Payne v. Tennessee, 501 U. S. 808, 827-828 (1991) (observing that reexamination of constitutional decisions is appropriate when those decisions have generated uncertainty and failed to provide clear guidance, because “correction through legis*951lative action is practically impossible” (internal quotation marks omitted)); Garcia v. San Antonio Metropolitan Transit Authority, 469 U. S. 528, 546-547, 557 (1985).

We have held that a liberty interest protected under the Due Process Clause of the Fourteenth Amendment will be deemed fundamental if it is “implicit in the concept of ordered liberty.” Palko v. Connecticut, 302 U. S. 319, 325 (1937). Three years earlier, in Snyder v. Massachusetts, 291 U. S. 97 (1934), we referred to a “principle of justice so rooted in the traditions and conscience of our people as to be ranked as fundamental.” Id., at 105; see also Michael H. v. Gerald D., 491 U. S. 110, 122 (1989) (plurality opinion) (citing the language from Snyder). These expressions are admittedly not precise, but our decisions implementing this notion of “fonda-mental” rights do not afford any more elaborate basis on which to base such a classification.

In construing the phrase “liberty” incorporated in the Due Process Clause of the Fourteenth Amendment, we have recognized that its meaning extends beyond freedom from physical restraint. In Pierce v. Society of Sisters, 268 U. S. 510 (1925), we held that it included a parent’s right to send a child to private school; in Meyer v. Nebraska, 262 U. S. 390 (1923), we held that it included a right to teach a foreign language in a parochial school. Building on these cases, we have held that the term “liberty” includes a right to marry, Loving v. Virginia, 388 U. S. 1 (1967); a right to procreate, Skinner v. Oklahoma ex rel. Williamson, 316 U. S. 535 (1942); and a right to use contraceptives, Griswold v. Connecticut, 381 U. S. 479 (1965); Eisenstadt v. Baird, 405 U. S. 438 (1972). But a reading of these opinions makes clear that they do not endorse any all-encompassing “right of privacy.”

In Roe v. Wade, the Court recognized a “guarantee of personal privacy” which “is broad enough to encompass a woman’s decision whether or not to terminate her pregnancy.” 410 U S., at 152-153. We are now of the view that, in terming this right fundamental, the Court in Roe read the earlier *952opinions upon which it based its decision much too broadly. Unlike marriage, procreation, and contraception, abortion “involves the purposeful termination of a potential life.” Harris v. McRae, 448 U. S. 297, 325 (1980). The abortion decision must therefore “be recognized as sui generis, different in kind from the others that the Court has protected under the rubric of personal or family privacy and autonomy.” Thornburgh v. American College of Obstetricians and Gynecologists, supra, at 792 (White, J., dissenting). One cannot ignore the fact that a woman is not isolated in her pregnancy, and that the decision to abort necessarily involves the destruction of a fetus. See Michael H. v. Gerald D., supra, at 124, n. 4 (To look “at the act which is assertedly the subject of a liberty interest in isolation from its effect upon other people [is] like inquiring whether there is a liberty interest in firing a gun where the case at hand happens to involve its discharge into another person’s body”).

Nor do the historical traditions of the American people support the view that the right to terminate one’s pregnancy is “fundamental.” The common - law which we inherited from England made abortion after “quickening” an offense. At the time of the adoption of the Fourteenth Amendment, statutory prohibitions or restrictions on abortion were commonplace; in 1868, at least 28 of the then-87 States and 8 Territories had statutes banning or limiting abortion. J. Mohr, Abortion in America 200 (1978). By the turn of the century virtually every State had a law prohibiting or restricting abortion on its books. By the middle of the present century, a liberalization trend had set in. But 21 of the restrictive abortion laws in effect in 1868 were still in effect in 1973 when Roe was decided, and an overwhelming majority of the States prohibited abortion unless necessary to preserve the life or health of the mother. Roe v. Wade, 410 U. S., at 139-140; id., at 176-177, n. 2 (Rehnquist, J., dissenting). On this record, it can scarcely be said that any deeply rooted tradition of relatively unrestricted abortion in our his*953tory supported the classification of the right to abortion as “fundamental” under the Due Process Clause of the Fourteenth Amendment.

We think, therefore, both in view of this history and of our decided cases dealing with substantive liberty under the Due Process Clause, that the Court was mistaken in Roe when it classified a woman's decision to terminate her pregnancy as a “fundamental right” that could be abridged only in a manner which withstood “strict scrutiny.” In so concluding, we repeat the observation made in Bowers v. Hardwick, 478 U. S. 186 (1986):

“Nor are we inclined to take a more expansive view of our authority to discover new fundamental rights imbedded in the Due Process Clause. The Court is most vulnerable and comes nearest to illegitimacy when it deals with judge-made constitutional law having little ■ or no cognizable roots in the language or design of the Constitution.” Id., at 194.

We believe that the sort of constitutionally imposed abortion code of the type illustrated by our decisions following Roe is inconsistent “with the notion of a Constitution cast in general terms, as ours is, and usually speaking in general principles, as ours does.” Webster v. Reproductive Health Services, 492 U. S., at 518 (plurality opinion). The Court in Roe reached too far when it analogized the right to abort a fetus to the rights involved in Pierce, Meyer, Loving, and Griswold, and thereby deemed the right to abortion fundamental.

II

The joint opinion of Justices O’Connor, Kennedy, and Souter cannot bring itself to say that Roe was correct as an original matter, but the authors are of the view that “the immediate question is not the soundness of Roe’s resolution of the issue, but the precedential force that must be aceorded to its holding.” Ante, at 871. Instead of claiming that Roe *954was correct as a matter of original constitutional interpretation, the opinion therefore contains an elaborate discussion of stare decisis. This discussion of the principle of stare decisis appears to be almost entirely dicta, because the joint opinion does not apply that principle in dealing with Roe. Roe decided that a woman had a fundamental right to an abortion. The joint opinion rejects that view. Roe decided that abortion regulations were to be subjected to “strict scrutiny” and could be justified only in the light of “compelling state interests.” The joint opinion rejects that view. Ante, at 872-873; see Roe v. Wade, supra, at 162-164. Roe analyzed abortion regulation under a rigid trimester framework, a framework which has guided this Court’s decisionmaking for 19 years. The joint opinion rejects that framework. Ante, at 873.

Stare decisis is defined in Black’s Law Dictionary as meaning “to abide by, or adhere to, decided cases.” Black’s Law Dictionary 1406 (6th ed. 1990). Whatever the “central holding” of Roe that is left after the joint opinion finishes dissecting it is surely not the result of that principle. While purporting to adhere to precedent, the joint opinion instead revises it. Roe continues to exist, but only in the way a storefront on a western movie set exists: a mere facade to give the illusion of reality. Decisions following Roe, such as Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416 (1983), and Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S. 747 (1986), are frankly overruled in part under the “undue burden” standard expounded in the joint opinion. Ante, at 881-884.

In our view, authentic principles of stare decisis do not require that any portion of the reasoning in Roe be kept intact. “Stare decisis is not... a universal, inexorable command,” especially in eases involving the interpretation of the Federal Constitution. Burnet v. Coronado Oil & Gas Co., 285 U. S. 393, 405 (1932) (Brandéis, J., dissenting). Erroneous decisions in such constitutional cases are uniquely durable, because correction through legislative action, save for *955constitutional amendment, is impossible. It is therefore our duty to reconsider constitutional interpretations that “depart] from a proper understanding” of the Constitution. Garcia v. San Antonio Metropolitan Transit Authority, 469 U. S., at 557; see United States v. Scott, 437 U. S. 82, 101 (1978) (“ ‘[I]n cases involving the Federal Constitution, . . . [t]he Court bows to the lessons of experience and the force of better reasoning, recognizing that the process of trial and error, so fruitful in the physical sciences, is appropriate also in the judicial function’” (quoting Burnet v. Coronado Oil & Gas Co., supra, at 406-408 (Brandeis, J., dissenting))); Smith v. Allwright, 321 U. S. 649, 665 (1944). Our constitutional watch does not cease merely because we have spoken before on an issue; when it becomes clear that a prior constitutional interpretation is unsound we are obliged to reexamine the question. See, e. g., West Virginia Bd. of Ed. v. Barnette, 319 U. S. 624, 642 (1943); Erie R. Co. v. Tompkins, 304 U. S. 64, 74-78 (1938).

The joint opinion discusses several stare decisis factors which, it asserts, point toward retaining a portion of Roe. Two of these factors are that the main “factual underpinning” of Roe has remained the same, and that its doctrinal foundation is no weaker now than it was in 1973. Ante, at 857-860. Of course, what might be called the basic facts which gave rise to Roe have remained the same — women become pregnant, there is a point somewhere, depending on medical technology, where a fetus becomes viable, and women give birth to children. But this is only to say that the same facts which gave rise to Roe will continue to give rise to similar cases. It is not a reason, in and of itself, why those cases must be decided in the same incorrect manner as was the first case to deal with the question. And surely there is no requirement, in considering whether to depart from stare decisis in a constitutional case, that a decision be more wrong now than it was at the time it was rendered. If that were true, the most outlandish constitutional decision could sur*956vive forever, based simply on the fact that it was no more outlandish later than it was when originally rendered.

Nor does the joint opinion faithfully follow this alleged requirement. The opinion frankly concludes that Roe and its progeny were wrong in failing to recognize that the State’s interests in maternal health and in the protection of unborn human life exist throughout pregnancy. Ante, at 871-873. But there is no indication that these components of Roe are any more incorrect at this juncture than they were at its inception.

The joint opinion also points to the reliance interests involved in this context in its effort to explain why precedent must be followed for precedent’s sake. Certainly it is true that where reliance is truly at issue, as in the case of judicial decisions that have formed the basis for private decisions, “[Considerations in favor of stare decisis are at their acme.” Payne v. Tennessee, 501 U. S., at 828. But, as the joint opinion apparently agrees, ante, at 855-856, any traditional notion of reliance is not applicable here. The Court today cuts back on the protection afforded by Roe, and no one claims that this action defeats any reliance interest in the disavowed trimester framework. Similarly, reliance interests would not be diminished were the Court to go further and acknowledge the full error of Roe, as “reproductive planning could take virtually immediate account. of” this action. Ante, at 856.

The joint opinion thus turns to what can only be described as an unconventional — and unconvincing — notion of reliance, a view based on the surmise that the availability of abortion since Roe has led to “two deeades of economic and social developments” that would be undercut if the error of Roe were recognized. Ante, at 856. The joint opinion’s assertion of this fact is undeveloped and totally conclusory. In fact, one cannot be sure to what economic and social developments the opinion is referring. Surely it is dubious to suggest that women have reached their “places in society” in *957reliance upon Roe, rather than as a result of their determination to obtain higher education and compete with men in the job market, and of society’s increasing recognition of their ability to fill positions that were previously thought to be reserved only for men. Ante, at 856.

In the end, having failed to put forth any evidence to prove any true reliance, the joint opinion’s argument is based solely on generalized assertions about the national psyche, on a belief that the people of this country have grown accustomed to the Roe decision over the last 19 years and have “ordered their thinking and living around” it. Ante, at 856, As an initial matter, one might inquire how the joint opinion can view the “central holding” of Roe as so deeply rooted in our constitutional culture, when it so casually uproots and disposes of that same decision’s trimester framework. Furthermore, at various points in the past, the same could have been said about this Court’s erroneous decisions that the Constitution allowed “separate but equal” treatment of minorities, see Plessy v. Ferguson, 163 U. S. 537 (1896), or that “liberty” under the Due Process Clause protected “freedom of contract,” see Adkins v. Children’s Hospital of District of Columbia, 261 U. S. 525 (1923); Lochner v. New York, 198 U. S. 45 (1905). The “separate but equal” doctrine lasted 58 years after Plessy, and Lochner’s protection of contractual freedom lasted 32 years. However, the simple fact that a generation or more had grown used to these major decisions did not prevent the Court from correcting its errors in those cases, nor should it prevent us from correctly interpreting the Constitution here. See Brown v. Board of Education, 347 U. S. 483 (1954) (rejecting the “separate but equal” doctrine); West Coast Hotel Co. v. Parrish, 300 U. S. 379 (1937) (overruling Adkins v. Children’s Hospital, supra, in upholding Washington’s minimum wage law).

Apparently realizing that conventional stare decisis principles do not support its position, the joint opinion advances a belief that retaining a portion of Roe is necessary to protect *958the “legitimacy” of this Court. Ante, at 861-869. Because the Court must take care to render decisions “grounded truly in principle,” and not simply as political and social compromises, ante, at 865, the joint opinion properly declares it to be this Court’s duty to ignore the public criticism and protest that may arise as a result of a decision. New would quarrel with this statement, although it may be doubted that Members of this Court, holding their tenure as they do during constitutional “good behavior,” are at all likely to be intimidated by such public protests.

But the joint opinion goes on to state that when the Court “resolve[s] the sort of intensely divisive controversy reflected in Roe and those rare, comparable cases,” its decision is exempt from reconsideration under established principles of stare decisis in constitutional eases. Ante, at 866. This is so, the joint opinion contends, because in those “intensely divisive” cases the Court has “eall[ed] the contending sides of a national controversy to end their national division by accepting a common mandate rooted in the Constitution,” and must therefore take special care not to be perceived as “surrendering] to political pressure” and continued opposition. Ante, at 866, 867. This is a truly novel principle, one which is contrary to both the Court’s historical practice and to the Court’s traditional willingness to tolerate criticism of its opinions. Under this principle, when the Court has ruled on a divisive issue, it is apparently prevented from overruling that decision for the sole reason that it was incorrect, unless opposition to the original decision has died away.

The first difficulty with this principle lies in its assumption that cases that are “intensely divisive” can be readily distinguished from those that are not. The question of whether a particular issue is “intensely divisive” enough to qualify for special protection is entirely subjective and dependent on the individual assumptions of the Members of this Court. In addition, because the Court’s duty is to ignore public opinion and criticism on issues that come before it, its Members are *959in perhaps the worst position to judge whether a decision divides the Nation deeply enough to justify such uncommon protection. Although many of the Court’s decisions divide the populace to a large degree, we have not previously on that account shied away from applying normal rules of stare decisis when urged to reconsider earlier decisions. Over the past 21 years, for example, the Court has overruled in whole or in part 34 of its previous constitutional decisions. See Payne v. Tennessee, supra, at 828-830, and n. 1 (listing eases).

The joint opinion picks out and discusses two prior Court rulings that it believes are of the “intensely divisive” variety, and concludes that they are of comparable dimension to Roe. Ante, at 861-864 (discussing Lochner v. New York, supra, and Plessy v. Ferguson, supra). It appears to us very odd indeed that the joint opinion chooses as benchmarks two cases in which the Court chose not to adhere to erroneous constitutional precedent, but instead enhanced its stature by acknowledging and correcting its error, apparently in violation of the joint opinion’s “legitimacy” principle. See West Coast Hotel Co. v. Parrish, supra; Brown v. Board of Education, supra. One might also wonder how it is that the joint opinion puts these, and not others, in the “intensely divisive” category, and how it assumes that these are the only two lines of cases of comparable dimension to Roe. There is no reason to think that either Plessy or Lochner produced the sort of public protest when they were decided that Roe did. There were undoubtedly large segments of the bench and bar who agreed with the dissenting views in those cases, but surely that cannot be what the Court means when it uses the term “intensely divisive,” or many other cases would have to be added to the list. In terms of public protest, however, Roe, so far as we know, was unique. But just as the Court should not respond to that sort of protest by retreating from the decision simply to allay the concerns of the protesters, it should likewise not respond by determining to adhere to the *960decision at all costs lest it seem to be retreating under fire. Public protests should not alter the normal application of stare decisis, lest perfectly lawful protest activity be penalized by the Court itself.

Taking the joint opinion on its own terms, we doubt that its distinction between Roe, on the one hand, and Plessy and Lochner, on the other, withstands analysis. The joint opinion acknowledges that the Court improved its stature by overruling Plessy in Brown on a deeply divisive issue. And our decision in West Coast Hotel, which overruled Adkins v. Children’s Hospital, supra, and Lochner, was rendered at a time when Congress was considering President Franklin Roosevelt’s proposal to “reorganize” this Court and enable him to name six additional Justices in the event that any Member of the Court over the age of 70 did not elect to retire. It is difficult to imagine a situation in which the Court would face more intense opposition to a prior ruling than it did at that time, and, under the general principle proclaimed in the joint opinion, the Court seemingly should have responded to this opposition by stubbornly refusing to reexamine the Lochner rationale, lest it lose legitimacy by appearing to “overrule under fire.” Ante, at 867.

The joint opinion agrees that the Court’s stature would have been seriously damaged if in Brown and West Coast Hotel it had dug in its heels and refused to apply normal principles of stare decisis to the earlier decisions. But the opinion contends that the Court was entitled to overrule Plessy and Lochner in those eases, despite the existence of opposition to the original decisions, only because both the Nation and the Court had learned new lessons in the interim. This is at best a feebly supported, post hoc rationalization for those decisions.

For example, the opinion asserts that the Court could justifiably overrule its decision in Lochner only because the Depression had convinced “most people” that constitutional protection of contractual freedom contributed to an economy *961that failed to protect the welfare of all. Ante, at 861. Surely the joint opinion does not mean to suggest that people saw this Court’s failure to uphold minimum wage statutes as the cause of the Great Depression! In any event, the Loch-ner Court did not base its rule upon the policy judgment that an unregulated market was fundamental to a stable economy; it simple believed, erroneously, that “liberty” under the Due Process Clause protected the “right to make a contract.” Lochner v. New York, 198 U. S., at 53. Nor is it the case that the people of this Nation only discovered the dangers of extreme laissez-faire economies because of the Depression. State laws regulating maximum hours and minimum wages were in existence well before that time. A Utah statute of that sort enacted in 1896 was involved in our decision in Holden v. Hardy, 169 U. S. 366 (1898), and other States followed suit shortly afterwards, see, e. g., Muller v. Oregon, 208 U. S. 412 (1908); Bunting v. Oregon, 243 U. S. 426 (1917). These statutes were indeed enacted because of a belief on the part of their sponsors that “freedom of contract” did not protect the welfare of workers, demonstrating that that belief manifested itself more than a generation before the Great Depression. Whether “most people” had come to share it in the hard times of the 1930’s is, insofar as anything the joint opinion advances, entirely speculative. The crucial failing at that time was not that workers were not paid a fair wage, but that there was no work available at any wage.

When the Court finally recognized its error in West Coast Hotel, it did not engage in the post hoc rationalization that the joint opinion attributes to it today; it did not state that Lochner had been based on an economic view that had fallen into disfavor, and that it therefore should be overruled. Chief Justice Hughes in his opinion for the Court simply recognized what Justice Holmes had previously recognized in his Lochner dissent, that “[t]he Constitution does not speak of freedom of contract.” West Coast Hotel Co. v. Parrish, 300 U. S., at 391; Lochner v. New York, supra, at 75 (Holmes, *962J., dissenting) (“[A] constitution is not intended to embody a particular economic theory, whether of paternalism and the organic relation of the citizen to the State or of laissez faire”). Although the Court did acknowledge in the last paragraph of its opinion the state of affairs during the then-current Depression, the theme of the opinion is that the Court had been mistaken as a matter of constitutional law when it embraced “freedom of contract” 32 years previously.

The joint opinion also agrees that the Court acted properly in rejecting the doctrine of “separate but equal” in Brown. In fact, the opinion lauds Brown in comparing it to Roe. Ante, at 867. This is strange, in that under the opinion’s “legitimacy” principle the Court would seemingly have been forced to adhere to its erroneous decision in Plessy because of its “intensely divisive” character. To us, adherence to Roe today under the guise of “legitimacy” would seem to resemble more closely adherence to Plessy on the same ground. Fortunately, the Court did not choose that option in Brown, and instead frankly repudiated Plessy. The joint opinion concludes that such repudiation was justified only because of newly discovered evidence that segregation had the effect of treating one race as inferior to another. But it can hardly be argued that this was not urged upon those who decided Plessy, as Justice Harlan observed in his dissent that the law at issue “puts the brand of servitude and degradation upon a large class of our fellow-citizens, our equals before the law.” Plessy v. Ferguson, 163 U. S., at 562. It is clear that the same arguments made before the Court in Brown were made in Plessy as well. The Court in Brown simply recognized, as Justice Harlan had recognized beforehand, that the Fourteenth Amendment does not permit racial segregation. The rule of Brown is not tied to popular opinion about the evils of segregation; it is a judgment that the Equal Protection Clause does not permit racial segregation, no matter whether the public might come to believe that it is beneficial. On that ground it stands, and on that ground *963alone the Court was justified in properly concluding that the Plessy' Court had erred.

There is also a suggestion in the joint opinion that the propriety of overruling a “divisive” decision depends in part on whether “iftost people” would now agree that it should be overruled. Either the demise of opposition or its progression to substantial popular agreement apparently is required to allow the Court to reconsider a divisive decision. How such agreement would be ascertained, short of a public opinion poll, the joint opinion does not say. But surely even the suggestion is totally at war with the idea of “legitimacy” in whose name it is invoked. The Judicial Branch derives its legitimacy, not from following public opinion, but from deciding by its best lights whether legislative enactments of the popular branches of Government comport with the Constitution. The doctrine of stare decisis is an adjunct of this duty, and should be no more subject to the vagaries of public opinion than is the basic judicial task.

There are other reasons why the joint opinion’s discussion of legitimacy is unconvincing as well. In assuming that the Court is perceived as “surrendering] to political pressure” when it overrules a controversial decision, ante, at 867, the joint opinion forgets that there are two sides to any controversy. The joint opinion asserts that, in order to protect its legitimacy, the Court must refrain from overruling a controversial decision lest it be viewed as favoring those who oppose the decision. But a decision to adhere to prior precedent is subject to the same criticism, for in such a case one can easily argue that the Court is responding to those who have demonstrated in favor of the original decision. The decision in Roe has engendered large demonstrations, including repeated marches on this Court and on Congress, both in opposition to and in support of that opinion. A decision either way on Roe can therefore be perceived as favoring one group or the other. But this perceived dilemma arises only if one assumes, as the joint opinion does, that the Court *964should make its decisions with a view toward speculative public perceptions. If one assumes instead, as the Court surely did in both Brown and West Coast Hotel, that the Court’s legitimacy is enhanced by faithful interpretation of the Constitution irrespective of public opposition, such self-engendered difficulties may be put to one side.

Roe is not this Court’s only decision to generate conflict. Our decisions in some recent capital eases, and in Bowers v. Hardwick, 478 U. S. 186 (1986), have also engendered demonstrations in opposition. The joint opinion’s message to such protesters appears to be that they must cease their activities in order to serve their cause, because their protests will only cement in place a decision which by normal standards of stare decisis should be reconsidered. Nearly a century ago, Justice David J. Brewer of this Court, in an article discussing criticism of its decisions, observed that “many criticisms may be, like their authors, devoid of good taste, but better all sorts of criticism than no criticism at all.” Justice Brewer on “The Nation’s Anchor,” 57 Albany L. J. 166, 169 (1898). This was good advice to the Court then, as it is today. Strong and often misguided criticism of a decision should not render the decision immune from reconsideration, lest a fetish for legitimacy penalize freedom of expression.

The end result of the joint opinion’s paeans of praise for legitimacy is the enunciation of a brand new standard for evaluating state regulation of a woman’s right to abortion— the “undue burden” standard. As indicated above, Roe v. Wade adopted a “fundamental right” standard under which state regulations could survive only if they met the requirement of “strict scrutiny.” While we disagree with that standard, it at least had a recognized basis in constitutional law at the time Roe was decided. The same cannot be said for the “undue burden” standard, which is created largely out of whole cloth by the authors of the joint opinion. It is a standard which even today does not command the support of a majority of this Gourt. And it will not, we believe, re-*965suit in the sort of “simple limitation,” easily applied, which the joint opinion anticipates. Ante, at 855. In sum, it is a standard which is not built to last.

In evaluating abortion regulations under that standard, judges will have to decide whether they place a “substantial obstacle” in the path of a woman seeking an abortion. • Ante, at 877. In that this standard is based even more on a judge’s subjective determinations than was the trimester framework, the standard will do nothing to prevent “judges from roaming at large in the constitutional field” guided only by their personal views. Griswold v. Connecticut, 381 U. S., at 502 (Harlan, J., concurring in judgment). Because the undue burden standard is plucked from nowhere, the question of what is a “substantial obstacle” to abortion will undoubtedly engender a variety of conflicting views. For example, in the very matter before us now, the authors of the joint opinion would uphold Pennsylvania’s 24-hour waiting period, concluding that a “particular burden” on some women is not a substantial obstacle. Ante, at 887. But the authors would at the same time strike down Pennsylvania’s spousal notice provision, after finding that in a “large fraction” of cases the provision will be a substantial obstacle. Ante, at 895. And, while the authors conclude that the informed consent provisions do not constitute an “undue burden,” Justice Stevens would hold that they do. Ante, at 920-922.

Furthermore, while striking down the spousal notice regulation, the joint opinion would uphold a parental consent restriction that certainly places very substantial obstacles in the path of a minor’s abortion choice. The joint opinion is forthright in admitting that it draws this distinction based on a policy judgment that parents will have the best interests of their children at heart, while the same is not necessarily true of husbands as to their wives. Ante, at 895. This may or may not be a correct judgment, but it is quintessentially a legislative one. The “undue burden” inquiry does not in any way supply the distinction between parental consent and *966spousal consent which the joint opinion adopts. Despite the efforts of the joint opinion, the undue burden standard presents nothing more workable than the trimester framework which it discards today. Under the guise of the Constitution, this Court will still impart its own preferences on the States in the form of a complex abortion code.

The sum of the joint opinion’s labors in the name of stare decisis and “legitimacy” is this: Roe v. Wade stands as a sort of judicial Potemkin Village, which may be pointed out to passers-by as a monument to the importance of adhering to precedent. But behind the facade, an entirely new method of analysis, without any roots in constitutional law, is imported to decide the constitutionality of state laws regulating abortion. Neither stare decisis nor “legitimacy” are truly served by such an effort.

We have stated above our belief that the Constitution does not subject state abortion regulations to heightened scrutiny. Accordingly, we think that the correct analysis is that set forth by the plurality opinion in Webster. A woman’s interest in having an abortion is a form of liberty protected by the Due Process Clause, but States may regulate abortion procedures in ways rationally related to a legitimate state interest. Williamson v. Lee Optical of Oklahoma, Inc., 348 U. S. 483, 491 (1955); cf. Stanley v. Illinois, 405 U. S. 645, 651-653 (1972). With this rule in mind, we examine each of the challenged provisions.

Ill

'A

Section 3205 of the Act imposes certain requirements related to the informed consent of a woman seeking an abortion. 18 Pa. Cons. Stat. §3205 (1990). Section 3205(a)(1) requires that the referring or performing physician must inform a woman contemplating an abortion of (i) the nature of the procedure and the risks and alternatives that a reasonable patient would find material; (ii) the fetus’ probable ges*967tational age; and (iii) the medical risks involved in carrying her pregnancy to term. Section 3205(a)(2) requires a physician or a nonphysician counselor to inform the woman that (i) the state health department publishes free materials describing the fetus at different stages and listing abortion alternatives; (ii) medical assistance benefits may be available for prenatal, childbirth, and neonatal care; and (iii) the child’s father is liable for child support. The Act also imposes a 24-hour waiting period between the time that the woman receives the required information and the time that the physician is allowed to perform the abortion. See Appendix to opinion of O’Connor, Kennedy, and Souter, JJ., ante, at 902-904.

This Court has held that it is certainly within the province of the States to require a woman’s voluntary and informed consent to an abortion. See Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 760. Here, Pennsylvania seeks to further its legitimate interest in obtaining informed consent by ensuring that each woman “is aware not only of the reasons for having an abortion, but also of the risks associated with an abortion and the availability of assistance that might make the alternative of normal childbirth more attractive than it might otherwise appear.” Id., at 798-799 (White, J., dissenting).

We conclude that this provision of the statute is rationally related to the State’s interest in assuring that a woman’s consent to an abortion be a fully informed decision.

Section 3205(a)(1) requires a physician to disclose certain information about the abortion procedure and its risks and alternatives. This requirement is certainly no large burden, as the Court of Appeals found that “the record shows that the clinics, without exception, insist on providing this information to women before an abortion is performed.” 947 F. 2d, at 703. We are of the view that this information “elearly is related to maternal health and to the State’s legitimate purpose in requiring informed consent.” Akron v. *968Akron Center for Reproductive Health, Inc., 462 U. S., at 446. An accurate description of the gestational age of the fetus and of the risks involved in carrying a child to term helps to further both those interests and the State’s legitimate interest in unborn human life. See id., at 445-446, n. 37 (required disclosure of gestational age of the fetus “certainly is not objectionable”). Although petitioners contend that it is unreasonable for the State to require that a physician, as opposed to a nonphysician counselor, disclose this information, we agree with the Court of Appeals that a State “may rationally decide that physicians are better qualified than counselors to impart this information and answer questions about the medical aspects of the available alternatives.” 947 F. 2d, at 704.

Section 3205(a)(2) compels the disclosure, by a physician or a counselor, of information concerning the availability of paternal child support and state-funded alternatives if the woman decides to proceed with her pregnancy. Here again, the Court of Appeals observed that “the record indicates that most clinics already require that a counselor consult in person with the woman about alternatives to abortion before the abortion is performed.” Id., at 704-705. And petitioners do not claim that the information required to be disclosed by statute is in any way false or inaccurate; indeed, the Court of Appeals found it to be “relevant, accurate, and noninflammatory.” Id., at 705. We conclude that this required presentation of “balanced information” is rationally related to the State’s legitimate interest in ensuring that the woman’s consent is truly informed, Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 830 (O’Connor, J., dissenting), and in addition furthers the State’s interest in preserving unborn life. That the information might create some uncertainty and persuade some women to forgo abortions does not lead to the conclusion that the Constitution forbids the provision of such information. Indeed, it only demonstrates that this information might *969very well make a difference, and that it is therefore relevant to a woman’s informed choice. Cf. id., at 801 (White, J., dissenting) (“[T]he ostensible objective of Roe v. Wade is not maximizing the number of abortions, but maximizing choice”). We acknowledge that in Thornburgh this Court struck down informed consent requirements similar to the ones at issue here. See id., at 760-764. It is clear, however, that while the detailed framework of Roe led to the Court’s invalidation of those informational requirements, they “would have been sustained under any traditional standard of judicial review, ... or for any other surgical procedure except abortion.” Webster v. Reproductive Health Services, 492 U. S., at 517 (plurality opinion) (citing Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 802 (White, J., dissenting); id., at 783 (Burger, C. J., dissenting)). In light of our rejection of Roe’s “fundamental right” approach to this subject, we do not regard Thornburgh as controlling.

For the same reason, we do not feel bound to follow this Court’s previous holding that a State’s 24-hour mandatory waiting period is unconstitutional. See Akron v. Akron Center for Reproductive Health, Inc., supra, at 449-451. Petitioners are correct that such a provision will result in delays for some women that might not otherwise exist, therefore placing a burden on their liberty. But the provision in no way prohibits abortions, and the informed consent and waiting period requirements do not apply in the ease of a medical emergency. See 18 Pa. Cons. Stat. §§ 3205(a), (b) (1990). We are of the view that, in providing time for reflection and reconsideration, the waiting period helps ensure that a woman’s decision to abort is a well-considered one, and reasonably furthers the State’s legitimate interest in maternal health and in the unborn life of the fetus. It “is surely a small cost to impose to ensure that the woman’s decision is well considered in light of its certain and irreparable conse-*970quenees on fetal life, and the possible effects on her own.” 462 U. S., at 474 (O’Connor, J., dissenting).

B

In addition to providing her own informed consent, before an unemaneipated woman under the age of 18 may obtain an abortion she must either furnish the consent of one of her parents, or must opt for the judicial procedure that allows her to bypass the consent requirement. Under the judicial bypass option, a minor can obtain an abortion if a state court finds that she is capable of giving her informed consent and has indeed given such consent, or determines that an abortion is in her best interests. Records of these court proceedings are kept confidential. The Act directs the state trial court to render a decision within three days of the woman’s application, and the entire procedure, including appeal to Pennsylvania Superior Court, is to last no longer than eight business days. The parental consent requirement does not apply in the case of a medical emergency. 18 Pa. Cons. Stat. §3206 (1990). See Appendix to opinion of O’Connor, Kennedy, and Souter, JJ., ante, at 904-906.

This provision is entirely consistent with this Court’s previous decisions involving parental consent requirements. See Planned Parenthood Assn. of Kansas City, Mo., Inc. v. Ashcroft, 462 U. S. 476 (1983) (upholding parental consent requirement with a similar judicial bypass option); Akron v. Akron Center for Reproductive Health, Inc., supra, at 439-440 (approving of parental consent statutes that include a judicial bypass option allowing a pregnant minor to “demonstrate that she is sufficiently mature to make the abortion decision herself or that, despite her immaturity, an abortion would be in her best interests”); Bellotti v. Baird, 443 U. S. 622 (1979).

We think it beyond dispute that a State “has a strong and legitimate interest in the welfare of its young citizens, whose immaturity, inexperience, and lack of judgment may some*971times impair their ability to exercise their rights wisely.” Hodgson v. Minnesota, 497 U. S., at 444 (opinion of Stevens, J.). A requirement of parental consent to abortion, like myriad other restrictions placed upon minors in other contexts, is reasonably designed to further this important and legitimate state interest. In our view, it is entirely “rational and fair for the State to conclude that, in most instances, the family will strive to give a lonely or even terrified minor advice that is both compassionate and mature.” Ohio v. Akron Center for Reproductive Health, 497 U. S., at 520 (opinion of Kennedy, J.); see also Planned Parenthood of Central Mo. v. Danforth, 428 U. S., at 91 (Stewart, J., concurring) (“There can be little doubt that the State furthers a constitutionally permissible end by encouraging an unmarried pregnant minor to seek the help and advice of her parents in making the very important decision whether or not to bear a child”). We thus conclude that Pennsylvania’s parental consent requirement should be upheld.

C

Section 3209 of the Act contains the spousal notification provision. It requires that, before a physician may perform an abortion on a married woman, the woman must sign a statement indicating that she has notified her husband of her planned abortion. A woman is not required to notify her husband if (1) her husband is not the father, (2) her husband, after diligent effort, cannot be located, (3) the pregnancy is the result of a spousal sexual assault that has been reported to the authorities, or (4) the woman has reason to believe that notifying her husband is likely to result in the infliction of bodily injury upon her by him or by another individual. In addition, a woman is exempted from the notification requirement in the case of a medical emergency. 18 Pa. Cons. Stat. §3209 (1990). See Appendix to opinion of O’Connor, Kennedy, and Souter, JJ., ante, at 908-909.

*972We first emphasize that Pennsylvania has not imposed a spousal consent requirement of the type the Court struck down in Planned Parenthood of Central Mo. v. Danforth, 428 U. S., at 67-72. Missouri’s spousal consent provision was invalidated in that case because of the Court’s view that it unconstitutionally granted to the husband “a veto power exercisable for any reason whatsoever or for no reason at all.” Id., at 71. But the provision here involves a much less intrusive requirement of spousal notification, not consent. Such a law requiring only notice to the husband “does not give any third party the legal right to make the [woman’s] decision for her, or to prevent her from obtaining an abortion should she choose to have one performed.” Hodgson v. Minnesota, supra, at 496 (Kennedy, J., concurring in judgment in part and dissenting in part); see H. L. v. Matheson, 450 U. S., at 411, n. 17. Danforth thus does not control our analysis. Petitioners contend that it should, however; they argue that the real effect of such a notice requirement is to give the power to husbands to veto a woman’s abortion choice. The District Court indeed found that the notification provision created a risk that some woman who would otherwise have an abortion will be prevented from having one. 947 F. 2d, at 712. For example, petitioners argue, many notified husbands will prevent abortions through physical force, psychological coercion, and other types of threats. But Pennsylvania has incorporated exceptions in the notice provision in an attempt to deal with these problems. For instance, a woman need not notify her husband if the pregnancy is the result of a reported sexual assault, or if she has reason to believe that she would suffer bodily injury as a result of the notification. 18 Pa. Cons. Stat. § 3209(b) (1990). Furthermore, because this is a facial challenge to the Act, it is insufficient for petitioners to show that the notification provision “might operate unconstitutionally under some conceivable set of circumstances.” United States v. Salerno, 481 U. S. 739, 745 (1987). Thus, it is not enough for petition*973ers to show that, in some “worst ease” circumstances, the notice provision will operate as a grant of veto power to husbands. Ohio v. Akron Center for Reproductive Health, 497 U. S., at 514. Because they are making a facial challenge to the provision, they must “show that no set of circumstances exists under which the [provision] would be valid.” Ibid. (internal quotation marks omitted). This they have failed to do.2

*974The question before us is therefore whether the spousal notification requirement rationally furthers any legitimate state interests. We conclude that it does. First, a husband’s interests in procreation within marriage and in the potential life of his unborn child are certainly substantial ones. See Planned Parenthood of Central Mo. v. Danforth, 428 U. S., at 69 (“We are not unaware of the deep and proper concern and interest that a devoted and protective husband has in his wife’s pregnancy and in the growth and development of the fetus she is carrying”); id., at 93 (White, J., concurring in part and dissenting in part); Skinner v. Oklahoma ex rel. Williamson, 316 U. S., at 541. The State itself has legitimate interests both in protecting these interests of the father and in protecting the potential life of the fetus, and the spousal notification requirement is reasonably related to advancing those state interests. By providing that a husband will usually know of his spouse’s intent to have an abortion, the provision makes it more likely that the husband will participate in deciding the fate of his unborn child, a possibility that might otherwise have been denied him. This participation might in some cases result in a decision to proceed with the pregnancy. As Judge Alito observed in his dissent below, “[t]he Pennsylvania legislature eould have rationally believed that some married women are initially inclined to obtain an abortion without their husbands’ knowledge because of perceived problems — such as economic constraints, future plans, or the husbands’ previously expressed *975opposition — that may be obviated by discussion prior to the abortion.” 947 F. 2d, at 726 (opinion concurring in part and dissenting in part).

The State also has a legitimate interest in promoting “the integrity of the marital relationship.” 18 Pa. Cons. Stat. § 3209(a) (1990). This Court has previously recognized “the importance of the marital relationship in our society.” Planned Parenthood of Central Mo. v. Danforth, supra, at 69. In our view, the spousal notice requirement is a rational attempt by the State to improve truthful communication between spouses and encourage collaborative decisionmaking, and thereby fosters marital integrity. See Labine v. Vincent, 401 U. S. 532, 538 (1971) (“[T]he power to make rules to establish, protect, and strengthen family life” is committed to the state legislatures). Petitioners argue that the notification requirement does not further any such interest; they assert that the majority of wives already notify their husbands of their abortion decisions, and the remainder have excellent reasons for keeping their decisions a secret. In the first ease, they argue, the law is unnecessary, and in the second ease it will only serve to foster marital discord and threats of harm. Thus, petitioners see the law as a totally irrational means of furthering whatever legitimate interest the State might have. But, in our view, it is unrealistic to assume that every husband-wife relationship is either (1) so perfect that this type of truthful and important communication will take place as a matter of course, or (2) so imperfect that, upon notice, the husband will react selfishly, violently, or contrary to the best interests of his wife. See Planned Parenthood of Central Mo. v. Danforth, supra, at 103-104 (Stevens, J., concurring in part and dissenting in part) (making a similar point in the context of a parental consent statute). The spousal notice provision will admittedly be unnecessary in some circumstances, and possibly harmful in others, but “the existence of particular cases in which a feature of a statute performs no function (or is even counterpro-*976duetive) ordinarily does not render the statute unconstitutional or even constitutionally suspect.” Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S., at 800 (White, J., dissenting). The Pennsylvania Legislature was in a position to weigh the likely benefits of the provision against its likely adverse effects, and presumably concluded, on balance, that the provision would be beneficial. Whether this was a wise decision or not, we cannot say that it was irrational. We therefore conclude that the spousal notice provision comports with the Constitution. See Harris v. McRae, 448 U. S., at 325-326 (“It is not the mission of this Court or any other to decide whether the balance of competing interests ... is wise social policy”).

D

The Act also imposes various reporting requirements. Section 3214(a) requires that abortion facilities file a report on each abortion performed. The reports do not include the identity of the women on whom abortions are performed, but they do contain a variety of information about the abortions. For example, each report must include the identities of the performing and referring physicians, the gestational age of the fetus at the time of abortion, and the basis for any medical judgment that a medical emergency existed. See 18 Pa. Cons. Stat. §§ 3214(a)(1), (5), (10) (1990). See Appendix to opinion of O’Connor, Kennedy, and Souter, JJ., ante, at 909-911. The District Court found that these reports are kept completely confidential. 947 F. 2d, at 716. We further conclude that these reporting requirements rationally farther the State’s legitimate interests in advancing the state of medical knowledge concerning maternal health and prenatal life, in gathering statistical information with respect to patients, and in ensuring compliance with other provisions of the Act.

Section 3207 of the Act requires each abortion facility to file a report with its name and address, as well as the names *977and addresses of any parent, subsidiary, or affiliated organizations. 18 Pa. Cons. Stat. § 3207(b) (1990). Section 3214(f) further requires each facility to file quarterly reports stating the total number of abortions performed, broken down by trimester. Both of these reports are available to the public only if the facility received state funds within the preceding 12 months. See Appendix to opinion of O’Connor, Kennedy, and Souter, JJ., ante, at 906,911. Petitioners do not challenge the requirement that facilities provide this information. They contend, however, that the forced public disclosure of the information given by facilities receiving public funds serves no legitimate state interest. We disagree. Records relating to the expenditure of public funds are generally available to the public under Pennsylvania law. See Pa. Stat. Ann., Tit. 65, §§66.1, 66.2 (Purdon 1959 and Supp. 1991-1992). As the Court of Appeals observed, “[w]hen a state provides money to a private commercial enterprise, there is a legitimate public interest in informing taxpayers who the funds are benefiting and what services the funds are supporting.” 947 F. 2d, at 718. These reporting requirements rationally further this legitimate state interest.

E

Finally, petitioners challenge the medical emergency exception provided for by the Act. The existence of a medical emergency exempts compliance with the Act’s informed consent, parental consent, and spousal notice requirements. See 18 Pa. Cons. Stat. §§ 3205(a), 3206(a), 3209(c) (1990). The Act defines a "medical emergency” as

“[t]hat condition which, on the basis of the physician’s good faith clinical judgment, so complicates the medical condition of a pregnant woman as to necessitate the immediate abortion of her pregnancy to avert her death or for which a delay will create serious risk of substantial *978and irreversible impairment of major bodily function.” §3203.

Petitioners argued before the District Court that the statutory definition was inadequate because it did not cover three serious conditions that pregnant women can suffer— preeelampsia, inevitable abortion, and prematurely ruptured membrane. The District Court agreed with petitioners that the medical emergency exception was inadequate, but the Court of Appeals reversed this holding. In construing the medical emergency provision, the Court of Appeals first observed that all three conditions do indeed present the risk of serious injury .or death when an abortion is not performed, and noted that the medical profession’s uniformly prescribed treatment for each of the three conditions is an immediate abortion. See 947 F. 2d, at 700-701. Finding that “[t]he Pennsylvania legislature did not choose the wording of its medical emergency exception in a vacuum,” the court read the exception as intended “to assure that compliance with its abortion regulations would not in any way pose a significant threat to the life or health of a woman.” Id., at 701. It thus concluded that the exception encompassed each of the three dangerous conditions pointed to by petitioners.

We observe that Pennsylvania’s present definition of medical emergency is almost an exact copy of that State’s definition at the time of this Court’s ruling in Thornburgh, one which the Court made reference to with apparent approval. 476 U. S., at 771 (“It is clear that the Pennsylvania Legislature knows how to provide a medieal-emergency exception when it chooses to do so”).3 We find that the interpretation *979of the Court of Appeals in these eases is eminently reasonable, and that the provision thus should be upheld. When a woman is faced with any condition that poses a “significant threat to [her] life or health,” she is exempted from the Act’s consent and notice requirements and may proceed immediately with her abortion.

IV

For the reasons stated, we therefore would hold that each of the challenged provisions of the Pennsylvania statute is consistent with the Constitution. It bears emphasis that our conclusion in this regard does not carry with it any necessary approval of these regulations. Our task is, as always, to decide only whether the challenged provisions of a law comport with the United States Constitution. If, as we believe, these do, their wisdom as a matter of public policy is for the people of Pennsylvania to decide.

Justice Scalia,

with whom The ChieF Justice, Justice White, and Justice Thomas join,

concurring in the judgment in part and dissenting in part.

My views on this matter are unchanged from those I set forth in my separate opinions in Webster v. Reproductive Health Services, 492 U. S. 490, 532 (1989) (opinion concurring in part and concurring in judgment), and Ohio v. Akron Center for Reproductive Health, 497 U. S. 502, 520 (1990) (Akron II) (concurring opinion). The States may, if they wish, permit abortion on demand, but the Constitution does not require them to do so. The permissibility of abortion, and the limitations upon it, are to be resolved like most important questions in our democracy: by citizens trying to persuade one another and then voting. As the Court acknowledges, “where reasonable people disagree the government can adopt one position or the other.”. Ante, at 851. The Court is correct in adding the qualification that this “assumes a state of affairs in which the choice does not intrude upon a protected liberty,” ibid. — but the crucial part of that quali*980fication is the penultimate word. A State’s choice between two positions on which reasonable people can disagree is constitutional even when (as is often the ease) it intrudes upon a “liberty” in the absolute sense. Laws against bigamy, for example — with which entire societies of reasonable people disagree — intrude upon men and women’s liberty to marry and live with one another. But bigamy happens not to be a liberty specially “protected” by the Constitution.

That is, quite simply, the issue in these eases: not whether the power of a woman to abort her unborn child is a “liberty” in the absolute sense; or even whether it is a liberty of great importance to many women. Of course it is both. The issue is whether it is a liberty protected by the Constitution of the United States. I am sure it is not. I reach that conclusion not because of anything so exalted as my views concerning the “concept of existence, of meaning, of the universe, and of the mystery of human life.” Ibid. Rather, I reach it for the same reason I reach the conclusion that bigamy is not constitutionally protected — because of two simple facts: (1) the Constitution says absolutely nothing about it, and (2) the longstanding traditions of American society have permitted it to be legally proscribed.1 Akron II, supra, at 520 (Scalia, J., concurring).

*981The Court destroys the proposition, evidently meant to represent my position, that “liberty” includes “only those practices, defined at the most specific level, that were protected against government interference by other rules of law when the Fourteenth Amendment was ratified,” ante, at 847 (citing Michael H. v. Gerald D., 491 U. S. 110, 127, n. 6 (1989) (opinion of Scalia, J.)). That is not, however, what Michael H. says; it merely observes that, in defining “liberty,” we may not disregard a specific, “relevant tradition protecting, or denying protection to, the asserted right,” ibid. But the Court does not wish to be fettered by any such limitations on its preferences. The Court’s statement that it is “tempting” to acknowledge the authoritativeness of tradition in order to “cur[b] the discretion of federal judges,” ante, at 847, is of course rhetoric rather than reality; no government official is “tempted” to place restraints upon his own freedom of action, which is why Lord Acton did not say “Power tends to purify.” The Court’s temptation is in the quite opposite and more natural direction — towards systematically eliminating checks upon its own power; and it succumbs.

Beyond that brief summary of the essence of my position, I will not swell the United States Reports with repetition of what I have said before; and applying the rational basis test, I would uphold the Pennsylvania statute in its entirety. I must, however, respond to a few of the more outrageous arguments in today’s opinion, which it is beyond human nature to leave unanswered. I shall discuss each of them under a quotation from the Court’s opinion to which they pertain.

“The inescapable fact is that adjudication of substantive due process claims may call upon the Court *982in interpreting the Constitution to exercise that same capacity which by tradition courts always have exercised: reasoned judgment.” Ante, at 849.

Assuming that the question before us is to be resolved at such a level of philosophical abstraction, in such isolation from the traditions of American society, as by simply applying “reasoned judgment,” I do not see how that could possibly have produced the answer the Court arrived at in Roe v. Wade, 410 U. S. 113 (1973). Today’s opinion describes the methodology of Roe, quite accurately, as weighing against the woman’s interest the State’s “ 'important and legitimate interest in protecting the potentiality of human life.’” Ante, at 871 (quoting Roe, supra, at 162). But “reasoned judgment” does not begin by begging the question, as Roe and subsequent eases unquestionably did by assuming that what the State is protecting is the mere “potentiality of human life.” See, e. g., Roe, supra, at 162; Planned Parenthood of Central Mo. v. Danforth, 428 U. S. 52, 61 (1976); Colautti v. Franklin, 439 U. S. 379, 386 (1979); Akron v. Akron Center for Reproductive Health, Inc., 462 U. S. 416, 428 (1983) (Akron I); Planned Parenthood Assn. of Kansas City, Mo., Inc. v. Ashcroft, 462 U. S. 476, 482 (1983). The whole argument of abortion opponents is that what the Court calls the fetus and what others call the unborn child is a human life. Thus, whatever answer Roe came up with after conducting its “balancing” is bound to be wrong, unless it is correct that the human fetus is in some critical sense merely potentially human. There is of course no way to determine that as a legal matter; it is in fact a value judgment. Some societies have considered newborn children not yet human, or the incompetent elderly no longer so.

The authors of the joint opinion, of course, do not squarely contend that Roe, v. Wade was a correct application of “reasoned judgment”; merely that it must be followed, because of stare decisis. Ante, at 853, 861, 871. But in their exhaustive discussion of all the factors that go into the determi*983nation of when stare decisis should be observed and when disregarded, they never mention “how wrong was the decision on its face?” Surely, if “[t]he Court’s power lies ... in its legitimacy, a product of substance and perception,” ante, at 865, the “substance” part of the equation demands that plain error be acknowledged and eliminated. Roe was plainly wrong — even on the Court’s methodology of “reasoned judgment,” and even more so (of course) if the proper criteria of text and tradition are applied.

The emptiness of the “reasoned judgment” that produced Roe is displayed in plain view by the fact that, after more than 19 years of effort by some of the brightest (and most determined) legal minds in the country, after more than 10 eases upholding abortion rights in this Court, and after dozens upon dozens of amicus briefs submitted in these and other eases, the best the Court can do to explain how it is that the word “liberty” must be thought to include the right to destroy human fetuses is to rattle off a collection of adjectives that simply decorate a value judgment and conceal a political choice. The right to abort, we are told, inheres in “liberty” because it is among “a person’s most basic decisions,” ante, at 849; it involves a “most intimate and personal choie[ej,” ante, at 851; it is “central to personal dignity and autonomy,” ibid.; it “originate^] within the zone of conscience and belief,” ante, at 852; it is “too intimate and personal” for state interference, ibid.; it reflects “intimate views” of a “deep, personal character,” ante, at 858; it involves “intimate relationships” and notions of “personal autonomy and bodily integrity,” ante, at 857; and it concerns a particularly “ ‘important decisio[n],’” ante, at 859 (citation omitted).2 But it is *984obvious to anyone applying “reasoned judgment” that the same adjectives can be applied to many forms of conduct that this Court (including one of the Justices in today’s majority, see Bowers v. Hardwick, 478 U. S. 186 (1986)) has held are not entitled to constitutional protection — because, like abortion, they are forms of conduct that have long been criminalized in American society. Those adjectives might be applied, for example, to homosexual sodomy, polygamy, adult incest, and suicide, all of which are equally “intimate” and “deep[ly] personal” decisions involving “personal autonomy and bodily integrity,” and all of which can constitutionally be proscribed because it is our unquestionable constitutional tradition that they are proseribable. It is not reasoned judgment that supports the Court’s decision; only personal predilection. Justice Curtis’s warning is as timely today as it was 135 years ago:

“[W]hen a strict interpretation of the Constitution, according to the fixed rules which govern the interpretation of laws, is abandoned, and the theoretical opinions of individuals are allowed to control its meaning, we have no longer a Constitution; we are under the government of individual men, who for the time being have power to declare what the Constitution is, according to their own views of what it ought to mean.” Dred Scott v. Sandford, 19 How. 393, 621 (1857) (dissenting opinion).

“Liberty finds no refuge in a jurisprudence of doubt.” Ante, at 844.

One might have feared to encounter this august and sonorous phrase in an opinion defending the real Roe v. Wade, rather than the revised version fabricated today by the au*985thors of the joint opinion. The shortcomings of Roe did not include lack of clarity: Virtually all regulation of abortion before the third trimester was invalid. But to come across this phrase in the joint opinion — which calls upon federal district judges to apply an “undue burden” standard as doubtful in application as it is unprincipled in origin — is really more than one should have to bear.

The joint opinion frankly concedes that the amorphous concept of “undue burden” has been inconsistently applied by the Members of this Court in the few brief years since that “test” was first explicitly propounded by Justice O’Connor in her dissent in Akron I, 462 U. S. 416 (1983). See ante, at 876.3 Because the three Justices now wish to “set forth a standard of general application,” the joint opinion announces that “it is important to clarify what is meant by an undue burden.” Ibid. I certainly agree with that, but I do not agree that the joint'opinion succeeds in the announced endeavor. To the contrary, its efforts at clarifica*986tion make elear only that the standard is inherently manipulable and will prove hopelessly unworkable in practice.

The joint opinion explains that a state regulation imposes an “undue burden” if it “has the purpose or effect of placing a substantial obstacle in the path of a woman seeking an abortion of a nonviable fetus.” Ante, at 877; see also ante, at 877-879. An obstacle is “substantial,” we are told, if it is “calculated^] [not] to inform the woman’s free choice, [but to] hinder it.” Ante, at 877.4 This latter statement cannot *987possibly mean what it says. Any regulation of abortion that is intended to advance what the joint opinion concedes is the State’s “substantial” interest in protecting unborn life will be “calculated [to] hinder” a decision to have an abortion. It thus seems more accurate to say that the joint opinion would uphold abortion regulations only if they do not unduly hinder the woman’s decision. That, of course, brings us right back to square one: Defining an “undue burden” as an “undue hindrance” (or a “substantial obstacle”) hardly “clarifies” the test. Consciously or not, the joint opinion’s verbal shell game will conceal raw judicial policy choices concerning what is “appropriate” abortion legislation.

The ultimately standardless nature of the “undue burden” inquiry is a reflection of the underlying fact that the concept has no principled or coherent legal basis. As The ChieF Justice points out, Roe’s strict-scrutiny standard “at least had a recognized basis in constitutional law at the time Roe was decided,” ante, at 964, while “[t]he same cannot be said for the ‘undue burden’ standard, which is created largely out of whole cloth by the authors of the joint opinion,” ibid. The joint opinion is flatly wrong in asserting that “our jurisprudence relating to all liberties save perhaps abortion has recognized” the permissibility of laws that do not impose an “undue burden.” Ante, at 873. It argues that the abortion right is similar to other rights in that a law “not designed to strike at the right itself, [but which] has the incidental effect of making it more difficult or more expensive to [exercise the right,]” is not invalid. Ante, at 874. I agree, indeed I have *988forcefully urged, that a law of general applicability which places only an incidental burden on a fundamental right does not infringe that right, see R. A. V. v. St. Paul, 505 U. S. 377, 389-390 (1992); Employment Div., Dept. of Human Resources of Ore. v. Smith, 494 U. S. 872, 878-882 (1990), but that principle does not establish the quite different (and quite dangerous) proposition that a law which directly regulates a fundamental right will not be found to violate the Constitution unless it imposes an “undue burden.” It is that, of course, which is at issue here: Pennsylvania has consciously and directly regulated conduct that our cases have held is constitutionally protected. The appropriate analogy, therefore, is that of a state law requiring purchasers of religious books to endure a 24-hour waiting period, or to pay a nominal additional tax of 10. The joint opinion cannot possibly be correct in suggesting that we would uphold such legislation on the ground that it does not impose a “substantial obstacle” to the exercise of First Amendment rights. The “undue burden” standard is not at all the generally applicable principle the joint opinion pretends it to be; rather, it is a unique concept created specially for these cases, to preserve some judicial foothold in this ill-gotten territory. In claiming otherwise, the three Justices show their willingness to place all constitutional rights at risk in an effort to preserve what they deem the “central holding in Roe.” Ante, at 873.

The rootless nature of the “undue burden”, standard, a phrase plucked out of context from our earlier abortion decisions, see n. 3, supra, is further reflected in the fact that the joint opinion finds it necessary expressly to repudiate the more narrow formulations used in Justice O’Connor’s earlier opinions. Ante, at 876-877. Those opinions stated that a statute imposes an “undue burden” if it imposes “absolute obstacles or severe limitations on the abortion decision,” Akron I, 462 U. S., at 464 (dissenting opinion) (emphasis added); see also Thornburgh v. American College of Obstetricians and Gynecologists, 476 U. S. 747, 828 (1986) (dissent*989ing opinion). Those strong adjectives are conspicuously missing from the joint opinion, whose authors have for some unexplained reason now determined that a burden is “undue” if it merely imposes a “substantial” obstacle to abortion decisions. See, e. g., ante, at 895, 901. Justice O’Connor has also abandoned (again without explanation) the view she expressed in Planned Parenthood Assn. of Kansas City, Mo., Inc. v. Ashcroft, 462 U. S. 476 (1983) (dissenting opinion), that a medical regulation which imposes an “undue burden” could nevertheless be upheld if it “reasonably relate[s] to the preservation and protection of maternal health,” id., at 505 (citation and internal quotation marks omitted). In today’s version, even health measures will be upheld only “if they do not constitute an undue burden,” ante, at 878 (emphasis added). Gone too is Justice O’Connor’s statement that “the State possesses compelling interests in the protection of potential human life . . . throughout pregnancy,” Akron I, supra, at 461 (dissenting opinion) (emphasis added); see also Ashcroft, supra, at 505 (O’Connor, J., concurring in judgment in part and dissenting in part); Thornburgh, supra, at 828 (O’Con-nor, J., dissenting); instead, the State’s interest in unborn human life is stealthily downgraded to a merely “substantial” or “profound” interest, ante, at 876, 878. (That had to be done, of course, since designating the interest as “compelling” throughout pregnancy would have been, shall we say, a “substantial obstacle” to the joint opinion’s determined effort to reaffirm what it views as the “central holding” of Roe. See Akron I, 462 U. S., at 420, n. 1.) And “viability” is no longer the “arbitrary” dividing line previously decried by Justice O’Connor in Akron I, id., at 461; the Court now announces that “the attainment of viability may continue to serve as the critical fact,” ante, at 860.5 It is difficult to *990maintain the illusion that we are interpreting a Constitution rather than inventing one, when we amend its provisions so breezily.

Because the portion of the joint opinion adopting and describing the undue burden test provides no more useful guidance than the empty phrases discussed above, one must turn to the 23 pages applying that standard to the present facts for further guidance. In evaluating Pennsylvania’s abortion law, the joint opinion relies extensively on the factual findings of the District Court, and repeatedly qualifies its conclusions by noting that they are contingent upon the record developed in these cases. Thus, the joint opinion would uphold the 24-hour waiting period contained in the Pennsylvania statute’s informed consent provision, 18 Pa. Cons. Stat. § 3205 (1990), because “the record evidence shows that in the vast majority of eases, a 24-hour delay does not create any appreciable health risk,” ante, at 885. The three Justices therefore conclude that “on the record before us,. .. we are not convinced that the 24-hour waiting period constitutes an undue burden.” Ante, at 887. The requirement that a doctor provide the information pertinent to informed consent would also be upheld because “there is no evidence on this record that [this requirement] would amount in practical terms to a substantial obstacle to a woman seeking- an abortion.” Ante, at 884. Similarly, the joint opinion would uphold the reporting requirements of the Act, §§3207, 3214, because “there is no . . . showing on the record before us” that these requirements constitute a “substantial obstacle” *991to abortion decisions. Ante, at 901. But at the same time the opinion pointedly observes that these reporting requirements may increase the costs of abortions and that “at some point [that fact] could become a substantial obstacle.” Ibid. Most significantly, the joint opinion’s conclusion that the spousal notice requirement of the Act, see § 3209, imposes an “undue burden” is based in large measure on the District Court’s “detailed findings of fact,” which the joint opinion sets out at great length, ante, at 888-891.

I do not, of course, have any objection to the notion that, in applying legal principles, one should rely only upon the facts that are contained in the record or that are properly subject to judicial notice.6 But what is remarkable about the joint opinion’s fact-intensive analysis is that it does not result in any measurable clarification of the “undue burden” standard. Rather, the approach of the joint opinion is, for the most part, simply to highlight certain facts in the record that apparently strike the three Justices as particularly significant in establishing (or refuting) the existence of an undue burden; after describing these facts, the opinion then simply announces that the provision either does or does not impose a “substantial obstacle” or an “undue burden.” See, e. g., ante, at 880, 884-885, 887, 893-894, 895, 901. We do not know whether the same conclusions could have been reached on a different record, or in what respects the record would have had to differ before an opposite conclusion would have been *992appropriate. The inherently standardless nature of this inquiry invites the district judge to give effect to his personal preferences about abortion. By finding and relying upon the right facts, he can invalidate, it would seem, almost any abortion restriction that strikes him as “undue” — subject, of course, to the possibility of being reversed by a court of appeals or Supreme Court that is as unconstrained in reviewing his decision as he was in making it.

To the extent I can discern any meaningful content in the “undue burden” standard as applied in the joint opinion, it appears' to be that a State may not regulate abortion in such a way as to reduce significantly its incidence. The joint opinion repeatedly emphasizes that an important factor in the “undue burden” analysis is whether the regulation “prevent[s] a significant number of women from obtaining an abortion,” ante, at 893; whether a “significant number of women . . . are likely to be deterred from procuring an abortion,” ante, at 894; and whether the regulation often “deters” women from seeking abortions, ante, at 897. We are not told, however, what forms of “deterrence” are impermissible or what degree of success in deterrence is too much to be tolerated. If, for example, a State required a woman to read a pamphlet describing, with illustrations, the facts of fetal development before she could obtain an abortion, the effect of such legislation might be to “deter” a “significant number of women” from procuring abortions, thereby seemingly allowing a district judge to invalidate it as an undue burden. Thus, despite flowery rhetoric about the State’s “substantial” and “profound” interest in “potential human life,” and criticism of Roe for undervaluing that interest, the joint opinion permits the State to pursue that interest only so long as it is not too successful. As Justice Blackmun recognizes (with evident hope), ante, at 926, the “undue burden” standard may ultimately require the invalidation of each provision upheld today if it can be shown, on a better record, that the State is too effectively “expressing] a pref-*993erenee for childbirth over abortion,” ante, at 883. Reason finds no refuge in this jurisprudence of confusion.

“While we appreciate the weight of the arguments . . . that Roe should be overruled, the reservations any of us may have in reaffirming the central holding of Roe are outweighed by the explication of individual liberty we have given combined with the force of stare decisis.” Ante, at 853.

The Court’s reliance upon stare decisis ean best be described as contrived. It insists upon the necessity of adhering not to all of Roe, but only to what it calls the “central holding.” It seems to me that stare decisis ought to be applied even to the doctrine of stare decisis, and I confess never to have heard of this new, keep-what-you-want-and-throwaway-the-rest version. I wonder whether, as applied to Marbury v. Madison, 1 Cranch 137 (1803), for example, the new version of stare decisis would be satisfied if we allowed courts to review the constitutionality of only those statutes that (like the one in Marbury) pertain to the jurisdiction of the courts.

I am certainly not in a good position to dispute that the Court has saved the “central holding” of Roe, since to do that effectively I would have to know what the Court has saved, which in turn would require me to understand (as I do not) what the “undue burden” test means. I must confess, however, that I have always thought, and I think a lot of other people have always thought, that the arbitrary trimester framework, which the Court today discards, was quite as central to Roe as the arbitrary viability test, which the Court today retains. It seems particularly ungrateful to carve the trimester framework out of the core of Roe, since its very rigidity (in sharp contrast to the utter indeterminability of the “undue burden” test) is probably the only reason the Court is able to say, in urging stare decisis, that Roe “has in no sense proven ‘unworkable/ ” ante, at 855. I suppose the *994Court is entitled to call a “central holding” whatever it wants to call a “central holding” — which is, come to think of it, perhaps one of the difficulties with this modified version of stare decisis. I thought I might note, however, that the following portions of Roe have not been saved:

• Under Roe, requiring that a woman seeking an abortion be provided truthful information about abortion before giving informed written consent is unconstitutional, if the information is designed to influence her choice. Thornburgh, 476 U. S., at 759-765; Akron I, 462 U. S., at 442-445. Under the joint opinion’s “undue burden” regime (as applied today, at least) such a requirement is constitutional. Ante, at 881-885.

• Under Roe, requiring that information be provided by a doctor, rather than by nonphysieian counselors, is unconstitutional. Akron I, supra, at 446-449. Under the “undue burden” regime (as applied today, at least) it is not. Ante, at 884-885.

• Under Roe, requiring a 24-hour waiting period between the time the woman gives her informed consent and the time of the abortion is unconstitutional. Akron I, supra, at 449-451. Under the “undue burden” regime (as applied today, at least) it is not. Ante, at 885-887.

• Under Roe, requiring detailed reports that include demographic data about each woman who seeks an abortion and various information about each abortion is unconstitutional. Thornburgh, supra, at 765-768. Under the “undue burden” regime (as applied today, at least) it generally is not. Ante, at 900-901.

“Where, in the performance of its judicial duties, the Court decides a case in such a way as to resolve the sort of intensely divisive controversy reflected in Roe ..., its decision has a dimension that the resolution of the normal case does not carry. It is the dimension present whenever the Court’s interpretation of the Constitution calls the contending sides of a *995national controversy to end their national division by accepting a common mandate rooted in the Constitution.” Ante, at 866-867.

The Court’s description of the place of Roe in the social history of the United States is unrecognizable. Not only did Roe not, as the Court suggests, resolve the deeply divisive issue of abortion; it did more than anything else to nourish it, by elevating it to the national level where it is infinitely more difficult to resolve. National politics were not plagued by abortion protests, national abortion lobbying, or abortion marches on Congress before Roe v. Wade was decided. Profound disagreement existed among our citizens over the issue — as it does over other issues, such as the death penalty — but that disagreement was being worked out at the state level. As with many other issues, the division of sentiment within each State was not as closely balanced as it was among the population of the Nation as a whole, meaning not only that more people would be satisfied with the results of state-by-state resolution, but also that those results would be more stable. Pr e-Roe, moreover, political compromise was possible.

Roe’s mandate for abortion on demand destroyed the compromises of the past, rendered compromise impossible for the future, and required the entire issue to be resolved'uniformly, at the national level. At the same time, Roe created a vast new class of abortion consumers and abortion proponents by eliminating the moral opprobrium that had attached to the act. (“If the Constitution guarantees abortion, how can it be bad?” — not an accurate line of thought, but a natural one.) Many favor all of those developments, and it is not for me to say that they are wrong. But to portray Roe as the statesmanlike “settlement” of a divisive issue, a jurisprudential Peace of Westphalia that is worth preserving, is nothing less than Orwellian. Roe fanned into life an issue that has inflamed our national polities in general, and has obscured with its smoke the selection of Justices to this Court *996in particular, ever since. And by keeping us in the abortion-umpiring business, it is the perpetuation of that disruption, rather than of any Pax Boeana, that the Court’s new majority decrees.

“[T]o overrule under fire . . . would subvert the Court’s legitimacy....

“... To all those who will be ... tested by following, the Court implicitly undertakes to remain steadfast .... The promise of constancy, once given, binds its maker for as long as the power to stand by the decision survives and ... the commitment [is not] obsolete....

“[The American people’s] belief in themselves as ... a people [who aspire to live according to the rule of law] is not readily separable from their understanding of the Court invested with the authority to decide their constitutional cases and speak before all others for their constitutional ideals. If the Court’s legitimacy should be undermined, then, so would the country be in its very ability to see itself through its constitutional ideals.” Ante, at 867-868.

The Imperial Judiciary lives. It is instructive to compare this Nietzschean vision of us uneleeted, life-tenured judges— leading a Yolk who will be “tested by following,” and whose very “belief in themselves” is mystically bound up in their “understanding” of a Court that “speak[s] before all others for their constitutional ideals” — with the somewhat more modest role envisioned for these lawyers by the Founders.

“The judiciary . . . has ... no direction either of the strength or of the wealth of the society, and can take no active resolution whatever. It may truly be said to have neither Force nor Will, but merely judgment....” The Federalist No. 78, pp. 393-394 (G. Wills ed. 1982).

Or, again, to compare this ecstasy of a Supreme Court in which there is, especially on controversial matters, no *997shadow of change or hint of alteration (“There is a limit to the amount of error that can plausibly be imputed to prior Courts,” ante, at 866), with the more democratic views of a more humble man:

“[Tjhe candid citizen must confess that if the policy of the Government upon vital questions affecting the whole people is to be irrevocably fixed by decisions of the Supreme Court,... the people will have ceased to be their own rulers, having to that extent practically resigned their Government into the hands of that eminent tribunal.” A. Lincoln, First Inaugural Address (Mar. 4,1861), reprinted in Inaugural Addresses of the Presidents of the United States, S. Doc. No. 101-10, p. 139 (1989).

It is particularly difficult, in the circumstances of the present decision, to sit still for the Court’s lengthy lecture upon the virtues of “constancy,” ante, at 868, of “remaining] steadfast,” ibid., and adhering to “principle,” ante, passim. Among the five Justices who purportedly adhere to Roe, at most three agree upon the principle that constitutes adherence (the joint opinion’s “undue burden” standard) — and that principle is inconsistent with Roe. See 410 U. S., at 154-156.7 To make matters worse, two of the three, in order thus to remain steadfast, had to abandon previously stated positions. See n. 4, supra; see supra, at 988-990. . It is beyond me how the Court expects these accommodations to be accepted “as grounded truly in principle, not as compromises with social and political pressures having, as such, no bearing on the principled choices that the Court is obliged to make.” Ante, at 865-866. The only principle the Court “adheres” *998to, it seems to me, is the principle that the Court must be seen as standing by Roe. That is not a principle of law (which is what I thought the Court was talking about), but a principle of Realpolitik — and a wrong one at that.

I cannot agree with, indeed I am appalled by, the Court’s suggestion that the decision whether to stand by an erroneous constitutional decision must be strongly influenced— against overruling, no less — by the substantial and continuing public opposition the decision has generated. The Court’s judgment that any other course would “subvert the Court’s legitimacy” must be another consequence of reading the error-filled history book that described the deeply divided country brought together by Roe. In my history book, the Court was covered with dishonor and deprived of legitimacy by Dred Scott v. Sandford, 19 How. 393 (1857), an erroneous (and widely opposed) opinion that it did not abandon, rather than by West Coast Hotel Co. v. Parrish, 300 U. S. 379 (1937), which produced the famous “switch in time” from the Court’s erroneous (and widely opposed) constitutional opposition to the social measures of the New Deal. (Both Dred Scott and one line of the cases resisting the New Deal rested upon the concept of “substantive due process” that the Court praises and employs today. Indeed, Dred Scott was “very possibly the first application of substantive due process in the Supreme Court, the original precedent for Lochner v. New York and Roe v. Wade.” D. Currie, The Constitution in the Supreme Court 271 (1985) (footnotes omitted).)

But whether it would “subvert the Court’s legitimacy” or not, the notion that we would decide a ease differently from the way we otherwise would have in order to show that we can stand firm against public disapproval is frightening. It is a bad enough idea, even in the head of someone like me, who believes that the text of the Constitution, and our traditions, say what they say and there is no fiddling with them. But when it is in the mind of a Court that believes the Con*999stitution has an evolving meaning, see ante, at 848; that the Ninth Amendment’s reference to “othe[r]” rights is not a disclaimer, but a charter for action, ibid.; and that the function of this Court is to “speak before all others for [the people’s] constitutional ideals” unrestrained by meaningful text or tradition — then the notion that the Court must adhere to a decision for as long as the decision faces “great opposition” and the Court is “under fire” acquires a character of almost ezarist arrogance. We are offended by these marchers who descend upon us, every year on the anniversary of Roe, to protest our saying that the Constitution requires what our society has never thought the Constitution requires. These people who refuse to be “tested by following” must be taught a lesson. We have no Cossacks, but at least we can stubbornly refuse to abandon an erroneous opinion that we might otherwise change — to show how little they intimidate us.

Of course, as The Chief Justice points out, we have been subjected to what the Court calls “‘political pressure’” by both sides of this issue. Ante, at 963. Maybe today’s decision not to overrule Roe will be seen as buckling to pressure from that direction. Instead of engaging in the hopeless task of predicting public perception — a job not for lawyers but for political campaign managers — the Justices should do what is legally right by asking two questions: (1) Was Roe correctly decided? (2) Has Roe succeeded in producing a settled body of law? If the answer to both questions is no, Roe should undoubtedly be overruled.

In truth, I am as distressed as the Court is — and expressed my distress several years ago, see Webster, 492 U. S., at 535 — about the “political pressure” directed to the Court: the marches, the mail, the protests aimed at inducing us to change our opinions. How upsetting it is, that so many of our citizens (good people, not lawless ones, on both sides of this abortion issue, and on various sides of other issues as well) think that we Justices should properly take into ac*1000count their views, as though we were engaged not in ascertaining an objective law but in determining some kind of social consensus. The Court would profit, I think, from giving less attention to the fact of this distressing phenomenon, and more attention to the cause of it. That cause permeates today’s opinion: a new mode of constitutional adjudication that relies not upon text and traditional practice to determine the law, but upon what the Court calls “reasoned judgment,” ante, at 849, which turns out to be nothing but philosophical predilection and moral intuition. All manner of “liberties,” the Court tells us, inhere in the Constitution and are enforceable by this Court — notjust those mentioned in the text or established in the traditions of our society. Ante, at 847-849. Why even the Ninth Amendment — which says only that “[t]he enumeration in the Constitution, of certain rights, shall not be construed to deny or disparage others retained by the people” — is, despite our contrary understanding for almost 200 years, a literally boundless source of additional, unnamed, unhinted-at “rights,” definable and enforceable by us, through “reasoned judgment.” Ante, at 848-849.

What makes all this relevant to the bothersome application of “political pressure” against the Court are the twin facts that the American people love democracy and the American people are not fools. As long as this Court thought (and the people thought) that we Justices were doing essentially lawyers’ work up here — reading text and discerning our society’s traditional understanding of that text — the public pretty much left us alone. Texts and traditions are facts to study, not convictions to demonstrate about. But if in reality our process of constitutional adjudication consists primarily of making value judgments; if we can ignore a long and clear tradition clarifying an ambiguous text, as we did, for example, five days ago in declaring unconstitutional invocations and benedictions at public high school graduation ceremonies, Lee v. Weisman, 505 U. S. 577 (1992); if, as I say, our pronouncement of constitutional law rests primarily on value *1001judgments, then a free and intelligent people’s attitude towards us can be expected to be (ought to be) quite different. The people know that their value judgments are quite as good as those taught in any law school — maybe better. If, indeed, the “liberties” protected by the Constitution are, as the Court says, undefined and unbounded, then the people should demonstrate, to protest that we do not implement their values instead of ours. Not only that, but confirmation hearings for new Justices should deteriorate into question- and-answer sessions in which Senators go through a list of their constituents’ most favored and most disfavored alleged constitutional rights, and seek the nominee’s commitment to support or oppose them. Value judgments, after all, should be voted on, not dictated; and if our Constitution has somehow accidently committed them to the Supreme Court, at least we can have a sort of plebiscite each time a new nominee to that body is put forward. Justice Blackmun not only regards this prospect with equanimity, he solicits it. Ante, at 943.

* * *

There is a poignant aspect to today’s opinion. Its length, and what might be called its epic tone, suggest that its authors believe they are bringing to an end a troublesome era in the history of our Nation and of our Court. “It is the dimension” of authority, they say, to “eal[l] the contending sides of national controversy to end their national division by accepting a common mandate rooted in the Constitution.” Ante, at 867.

There comes vividly to mind a portrait by Emanuel Leutze that hangs in the Harvard Law School: Roger Brooke Taney, painted in 1859, the 82d year of his life, the 24th of his Chief Justiceship, the second after his opinion in Dred Scott. He is all in black, sitting in a shadowed red armchair, left hand resting upon a pad of paper in his lap, right hand hanging limply, almost lifelessly, beside the inner arm of the chair. He sits facing the viewer and staring straight out. There *1002seems to be on his face, and in his deep-set eyes, an expression of profound sadness and disillusionment. Perhaps he always looked that way, even when dwelling upon the happiest of thoughts. But those of us who know how the lustre of his great Chief Justiceship came to be eclipsed by Dred Scott cannot help believing that he had that ease — its already apparent consequences for the Court and its soon-to-be-played-out consequences for the Nation — burning on his mind. I expect that two years earlier he, too, had thought himself “eall[ing] the contending sides of national controversy to end their national division by accepting a common mandate rooted in the Constitution.”

It is no more realistic for us in this litigation, than it was for him in that, to think that an issue of the sort they both involved — an issue involving life and death, freedom and subjugation — can be “speedily and finally settled” by the Supreme Court, as President James Buchanan in his inaugural address said the issue of slavery in the territories would be. See Inaugural Addresses of the Presidents of the United States, S. Doc. No. 101-10, p. 126 (1989). Quite to the contrary, by foreclosing all democratic outlet for the deep passions this issue arouses, by banishing the issue from the political forum that gives all participants, even the losers, the satisfaction of a fair hearing and an honest fight, by continuing the imposition of a rigid national rule instead of allowing for regional differences, the Court merely prolongs and intensifies the anguish.

We should get out of this area, where we have no right to be, and where we do neither ourselves nor the country any good by remaining.

GRUENDER, Circuit Judge.

The Governor and Attorney General of South Dakota (“the State”), along with two intervening crisis pregnancy centers and two of their personnel (collectively “Intervenors”), appeal the district court’s permanent injunction barring enforcement of a South Dakota statute requiring the disclosure to patients seeking abortions of an “[ijncreased risk of suicide ideation and suicide,” see S.D.C.L. § 34-23A-10.1(l)(e)(ii) (“suicide advisory”), and the underlying grant of summary judgment in favor of Planned Parenthood of Minnesota, North Dakota, South Dakota and its medical director Dr. Carol E. Ball (collectively “Planned Parenthood”) that this advisory would unduly burden abortion rights and would violate physicians’ First Amendment right to be free from compelled speech. For the reasons discussed below, we reverse.

I.

In 2005, South Dakota enacted House Bill 1166 (“the Act”), amending the requirements for obtaining informed consent to an abortion as codified in S.D.C.L. § 34-23A-10.1. Section 7 of the Act requires physicians, in the course of obtaining informed consent, to provide certain information to the patient seeking an abortion. In June 2005, Planned Parenthood sued to prevent the Act from taking effect, contending that several of its provisions constituted an undue burden on abortion rights and facially violated patients’ and physicians’ free speech rights, while other provisions were unconstitutionally vague. After the district court preliminarily enjoined the Act and a divided panel of this court affirmed, this court sitting en banc vacated the preliminary injunction and remanded for further proceedings. See Planned Parenthood Minn., N.D., S.D. v. Rounds, 530 F.3d 724 (8th Cir.2008) (en banc).

On remand, the parties filed cross-motions for summary judgment with respect to the challenged provisions. The district court ruled that a biological disclosure, see §§ 34-23A-10.1(l)(b), 34-23A-R4), and a medical emergency exception, see § 34-23A-10.1, were facially sound with respect to the First Amendment and imposed no undue burden, while disclosures regarding the protected relationship between the patient and the unborn child, see § 34-23A-10.1(l)(c), (d), and the suicide advisory, see § 34-23A-10.1(l)(e)(ii), failed to meet both constitutional requirements. The district court also held that a requirement to disclose “all known medical risks of the procedure,” see § 34-23A-10.1(l)(e), was not *893unconstitutionally vague, but that a requirement to disclose “statistically significant risk factors,” see id,., was.

Planned Parenthood appealed the district court’s decision on the biological disclosure and the “all known medical risks” disclosure, while the State and Intervenors appealed the district court’s decision on the relationship disclosures and the suicide advisory. A panel of this court affirmed unanimously with respect to the biological disclosure and the “all known medical risks” disclosure, reversed unanimously with respect to the relationship disclosures, and affirmed in a divided decision as to the suicide advisory. See Planned Parenthood Minn., N.D., S.D. v. Rounds, 653 F.3d 662 (8th Cir.2011). We granted this rehearing en banc solely on the issue of the suicide advisory.1

II.

We review a grant of summary judgment de novo. Missouri ex rel. Nixon v. Am. Blast Fax, Inc., 323 F.3d 649, 653 (8th Cir.2003). In addition, we review constitutional challenges and questions of statutory interpretation de novo. McDermott v. Royal, 613 F.3d 1192, 1193 (8th Cir.2010) (per curiam).

Planned Parenthood contends that requiring a physician to present the suicide advisory imposes an undue burden on abortion rights and violates the free speech rights of the physician. “[W]hen the government requires [as part of the informed consent process] ... the giving of truthful, nonmisleading information about the nature of the procedure, the attendant health risks and those of childbirth,” and other information broadly relevant to the decision to have an abortion, it does not impose an undue burden on abortion rights, even if the disclosure “might cause the woman to choose childbirth over abortion.” Planned Parenthood of Se. Penn. v. Casey, 505 U.S. 833, 882-83, 112 S.Ct. 2791, 120 L.Ed.2d 674 (1992). Moreover, “the physician’s First Amendment rights not to speak are implicated, but only as part of the practice of medicine, subject to reasonable licensing and regulation by the State.” Id. at 884, 112 S.Ct. 2791 (citations omitted). Thus, with respect to First Amendment concerns, “while the State cannot compel an individual simply to speak the State’s ideological message, it can use its regulatory authority to require a physician to provide truthful, non-misleading information relevant to a patient’s decision to have an abortion, even if that information might also encourage the patient to choose childbirth over abortion.” Rounds, 530 F.3d at 734-35; accord Tex. Med. Providers Performing Abortion Servs. v. Lakey, 667 F.3d 570, 576-77 (5th Cir.2012).

In short, to succeed on either its undue burden or compelled speech claims, Planned Parenthood must show that the disclosure at issue “is either untruthful, misleading or not relevant to the patient’s decision to have an abortion.” Rounds, 530 F.3d at 735. To evaluate the constitutional merits of the suicide advisory, we will examine first what disclosure actually is required, second whether that disclosure is truthful, and third whether it is non-misleading and relevant to the patient’s decision to have an abortion.

III.

Section 34-23A-10.1 requires a physician seeking to perform an abortion to present to the patient:

(1) A statement in writing providing the following information:

*894(e) A description of all known medical risks of the procedure and statistically significant risk factors to which the pregnant woman would be subjected, including:

(i) Depression and related psychological distress;

(ii) Increased risk of suicide ideation and suicide;

Planned Parenthood argues, and the district court agreed, that subsection (ii) must be construed to require a disclosure of a conclusive causal link between abortion and suicide. See Planned Parenthood Minn., N.D., S.D. v. Rounds, 650 F.Supp.2d 972, 982 (D.S.D.2009). However, no language in subsection (ii), or in the heading of section 10.1(l)(e), refers to such a causal link. “The intent of a statute is determined from what the legislature said, rather than what the courts think it should have said, and the court must confíne itself to the language used.” Langdeau v. Langdeau, 751 N.W.2d 722, 727 (S.D.2008) (quoting US W. Commc’ns, Inc. v. Pub. Utils. Comm’n, 505 N.W.2d 115, 123 (S.D.1993)).

Here, the language actually used by the legislature — “medical risks,” “statistically significant risk factors,” “[i]n-creased risk” — denotes risk in a medical context. Moreover, while the heading of subsection (e) refers broadly to “all known medical risks of the [abortion] procedure ... including” those listed in its subsections, the suicide advisory is the only subsection to further incorporate the more precise phrase “[¡Increased risk.” See § 34-23A-10.1(l)(e)(ii). Therefore, we must presume that the term “increased risk” has a more precise meaning than the umbrella term “risk” by itself. See Maynard v. Heeren, 563 N.W.2d 830, 835 (S.D.1997) (“[N]o wordage should be found to be surplus. No provision can be left without meaning. If possible, effect should be given to every part and every word.” (quoting Cummings v. Mickelson, 495 N.W.2d 493, 500 (S.D.1993))); see also FCC v. AT & T Inc., — U.S.-, 131 S.Ct. 1177, 1183, 179 L.Ed.2d 132 (2011) (recognizing that, in construing a statute, “two words together may assume a more particular meaning than those words in isolation”). The term “increased risk” is not defined in the statute, and it has more than one reasonable definition in the medical field. South Dakota law requires that such a term “must be construed according to its accepted usage, and a strained, unpractical or absurd result is to be avoided.” Peters v. Spearfish ETJ Planning Comm’n, 567 N.W.2d 880, 885 (S.D.1997).

As a result, the disclosure actually required by the suicide advisory depends upon the accepted usage of the term “increased risk” in the relevant medical field. We turn to the medical literature and expert evidence in the record to discern the accepted usage of the term “increased risk” in the applicable medical context, with an eye towards whether that accepted usage necessarily implies proof of causation.

The peer-reviewed medical literature in the record on the topic of suicide and abortion consistently uses the term “increased risk” to refer to a relatively higher probability of an adverse outcome in one group compared to other groups — that is, to “relative risk.” See Stedman’s Medical Dictionary 1701 (28th ed. 2006) (defining relative risk as “the ratio of the r[isk] of disease among those exposed to a r[isk] factor to the r[isk] among those not exposed”). For example, one study compared the rate of suicide for women who had received an induced abortion with the rates of suicide for two other groups, women who had given birth and women who had miscarried. See Ex. 60, Mika Gissler *895et al., Suicides After Pregnancy in Finland, 1987-94, 313 Brit. Med. J. 1431, 1432 (1996), ECF No. 172-3.2 That study characterized its finding of a vastly higher suicide rate for women who received an induced abortion as “an increased risk of suicide.” Id. at 1434. Another study compared the rate of, inter alia, suicide ideation in women who had received an induced abortion with the rates for women who had given birth and for women who had not become pregnant. See Ex. 61, David M. Fergusson et al., Abortion in Young Women and Subsequent Mental Health, 47 J. Child Psychol. & Psychiatry 16, 19 (2006), ECF No. 172-4. That study concluded, “Certainly in this study, those young women who had abortions appeared to be at moderately increased risk of both concurrent and subsequent mental health problems when compared with equivalent groups of pregnant or non-pregnant peers.” Id. at 23 (emphasis added).

The discussion of risk in the medical context provided by Intervenors’ expert also supports the conclusion that the term “increased risk” refers to the comparison of two groups, or relative risk:

Assessment of degree of risk is often expressed in terms of absolute risk, which relates to the chance of developing a disease over a time-period (e.g., a 10% lifetime risk of suicide) or in terms of relative risk, which is a comparison of the probability of an adverse outcome in two groups. For example, abortion would be considered an increased risk for suicide if the relative risk is significantly higher for women who abort compared to women who give birth or never have children.

Coleman Decl. ¶ 6, Jul. 6, 2006, ECF No. 189 (emphases added). Based on the “accepted usage” of the term in the relevant field, Peters, 567 N.W.2d at 885, the term “increased risk” in subsection (ii) indicates that the “relative risk” definition is the one intended by the legislature for the suicide advisory.

Noticeably absent from the contextual definition of “increased risk” is a requirement for conclusive proof of causation. This stands to reason, because, as explained by the Intervenors’ expert:

When examining complex human psychological and physical health outcomes, such as depression and suicidal behavior, identification of a single, precise causal mechanism applicable to all situations is not possible----

Given this inherent complexity, sound epidemiological evidence is nevertheless derived by identifying those variables which are most strongly linked with adverse mental or physical health outcomes for large groups of individuals.

Coleman Decl. ¶¶ 5-6, Jul. 6, 2006. While such evidence of relative risk eventually may prove direct causation as further experiments rule out plausible competing explanations, see id. at ¶ 9, conclusive proof of causation is not required in order for the identification of a medical risk.

Even the evidence upon which Planned Parenthood heavily relies is consistent with the “relative risk” definition of “increased risk,” with no requirement for proof of causation. For example, the report of the American Psychological Association’s (“APA”) Task Force on Mental Health and Abortion, Branson Decl. Ex. A, Sept. 8, 2008, ECF Nos. 283-3, 283-4 (hereinafter “APA Report”), decries the “tendency to confuse a risk and a cause” as a “logical fallacy.” APA Report at 31. As another example, Planned Parenthood submitted into the record a letter to a medical *896journal from one of the researchers mentioned above. While the researcher emphasized that his studies linking suicide and abortion did not prove causation, he resolutely reiterated his finding of “increased risk.” Mika Gissler et al., Letter to the Editor: Pregnancy-Related Violent Deaths, 27 Scand. J. Pub. Health 1:54, 55 (1999), ECF No. 206-10. It would be nonsensical for those in the field to distinguish a relationship of “increased risk” from one of causation if the term “risk” itself was equivalent to causation.

In the face of this extensive evidence of the accepted usage of the term “increased risk,” Planned Parenthood makes two arguments as to why the suicide advisory should be read to require a disclosure of proof of causation. First, it argues that the statute refers to the “increased risk of suicide ideation and suicide” as a risk “to which the pregnant woman would be subjected ” by the abortion procedure, see § 34-23A-10.1(l)(e) (emphasis added), implying that the abortion procedure directly subjects the patient to, or causes, the result. A relevant rule of statutory construction, however, holds that “a limiting clause or phrase ... should ordinarily be read as modifying only the noun or phrase that it immediately follows.” Barnhart v. Thomas, 540 U.S. 20, 26, 124 S.Ct. 376, 157 L.Ed.2d 333 (2003). Under that rule, the phrase “to which the pregnant woman would be subjected” modifies only the immediately preceding phrase “statistically significant risk factors” (which is not at issue here), not the phrase “all known medical risks of the procedure” (of which the “increased risk of suicide ideation and suicide” is a listed example). See § 34-23A-10.1(l)(e).

Moreover, even if the phrase “to which the pregnant woman would be subjected” is construed to modify the “increased risk” language, it would not advance Planned Parenthood’s argument because the result to which the pregnant woman would be subjected is the increased risk. In other words, the abortion procedure causes the patient to become a member of a group for which an increased risk is documented relative to other groups. This does not imply proof that the abortion procedure directly causes the adverse outcome in those cases where the risk materializes. There is a very real difference between (1) a statement that an action places an individual at an increased risk for an adverse outcome, and (2) a statement that, if the individual experiences the adverse outcome, the action will have been the direct cause.3

Second, Planned Parenthood relies on the “established principle of statutory construction that, where the wording of an act is changed by amendment, it is evidential of an intent that the words shall have a different construction.” Lewis & Clark Rural Water Sys., Inc. v. Seeba, 709 N.W.2d 824, 831 (S.D.2006) (quoting S.D. Subsequent Injury Fund v. Federated Mut. Ins., Inc., 605 N.W.2d 166, 170 (S.D.2000)). The informed-consent statute in *897effect prior to the Act required the disclosure of “[t]he particular medical risks associated with the particular abortion procedure to be employed including, when medically accurate, the risks of infection, hemorrhage, danger to subsequent pregnancies, and infertility.” S.D.C.L. § 34-23A-10.1(l)(b) (2004) (emphasis added). The Act expanded this subject matter into a new, four-part subsection:

(e) A description of all known medical risks of the procedure and statistically significant risk factors to which the pregnant woman would be subjected, including:

(i) Depression and related psychological distress;

(ii) Increased risk of suicide ideation and suicide;

(in) A statement setting forth an accurate rate of deaths due to abortions, including all deaths in which the abortion procedure was a substantial contributing factor;

(iv) All other known medical risks to the physical health of the woman, including the risk of infection, hemorrhage, danger to subsequent pregnancies, and infertility!)]

§ 34-23A-10.1(l)(e) (2005). Because this provision as amended by the Act no longer includes the word “associated,” Planned Parenthood asks us to conclude that the legislature intended the term “increased risk” to imply proof of causation, rather than that the procedure and the adverse outcome are merely “associated” by a correlative relationship such as relative risk.

We certainly agree that the amendments to the medical-risks provision are “evidential of an intent that the words shall have a different construction,” Lewis & Clark Rural Water Sys., 709 N.W.2d at 831 (quoting S.D. Subsequent Injury Fund, 605 N.W.2d at 170), but in this case that different construction does not hinge on the removal of one word. Instead, the Act effects essentially a complete rewriting of the former § 34-23A-10.1(l)(b) (2004), removing thirteen of the original twenty-eight words and adding seventy new words, including an entirely new introduction requiring a description of “all known medical risks” and a listing of three new specific areas of concern in subsections (i)(iii). Taken as a whole, these sweeping changes to the language of the provision express the legislature’s intent to address a much broader range of specific medical risks in the required disclosure, not to implicitly sever the term “increased risk” from its accepted usage in the medical field. See Lewis & Clark Rural Water Sys., 709 N.W.2d at 830 (“[T]he true intent of the legislature in enacting laws ... is ascertained primarily from the language employed in the statute.” (quoting Sanford v. Sanford, 694 N.W.2d 283, 287 (2005))). Indeed, where only fifteen words of original language remain in an amended provision of eighty-five words, ascribing such an effect to the removal of a single word would go far beyond any use of the cited rule of statutory construction of which we are aware. See, e.g., S.D. Subsequent Injury Fund, 605 N.W.2d at 170-71 (applying the statutory-amendment rule of construction to a seventy-one-word statute to which four new words were added and two were changed).

Finally, even if the language of the suicide advisory also reasonably could be construed to require a disclosure of a causal link, we would be faced with “varying constructions of the South Dakota statute, ‘by [one] of which grave and doubtful constitutional questions arise and by [the other] of which such questions are avoided.’” Rounds, 653 F.3d at 669 (quoting United States v. Adler, 590 F.3d 581, 583 (8th Cir.2009)). In such a situation, our “duty is to adopt the latter,” id. (quoting Adler, 590 F.3d at 583), and “[t]his is *898especially so since ‘[i]n evaluating a facial challenge to a state law, a federal court must ... consider any limiting construction that a, state ... enforcement agency has proffered,’ ” id. (quoting Kolender v. Lawson, 461 U.S. 352, 355, 103 S.Ct. 1855, 75 L.Ed.2d 903 (1983)). As a result, we would be called to apply the “relative risk” construction of increased risk over a construction that required disclosure of a causal link.

To summarize, in subsection (ii), the legislature expressly required the disclosure of an “increased risk,” not a causal link. Based on the accepted usage of the term “increased risk” in the relevant medical field, the usage of that term in the context of § 34 — 23A—10.1(l)(e)(ii) does not imply a disclosure of a causal relationship. Instead, subsection (ii) requires a disclosure simply that the risk of suicide and suicide ideation is higher among women who abort compared to women in other relevant groups, such as women who give birth or do not become pregnant.

IV.

With regard to whether the required disclosure is truthful, see Rounds, 530 F.3d at 735, the State submitted into the record numerous studies published in peer-reviewed medical journals that demonstrate a statistically significant correlation between abortion and suicide. The studies were published in respected, peer-reviewed journals such as the Obstetrical and Gynecological Survey, the British Medical Journal, the Journal of Child Psychology and Psychiatry, the Southern Medical Journal, and the European Journal of Public Health, and there is no indication that the peer-review process was compromised for the studies at issue. See Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 593-94, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993) (“The fact of publication (or lack thereof) in a peer reviewed journal ... [is] a relevant, though not dis-positive, consideration in assessing the scientific validity of a particular technique or methodology on which an opinion is premised” because “submission to the scrutiny of the scientific community ... increases the likelihood that substantive flaws in methodology will be detected.”).

Planned Parenthood argues that these studies do not examine the correlation between abortion and suicide in sufficient detail to prove a causal link (as discussed in more detail in Part V), but, as we concluded above, the suicide advisory does not require disclosure of a causal link. With regard to the accuracy of the correlation itself, there is nothing in the record to suggest that the underlying data or calculations in any of these studies are flawed. For example, Planned Parenthood’s own expert, Dr. Nada Stotland, admitted that one of the studies, which determined a suicide rate after abortion of 31.9 per 100,-000 as compared to a suicide rate after live birth of 5.0 per 100,000, “indicates an association; not causation, but an association” between abortion and suicide. Stotland Dep. 283:22-284:9, ECF No. 152-12.4 When asked if she had “any quarrel with the validity of that association,” Dr. Stotland replied that she did not. Id. at 284:11-13.

Based on the record, the studies submitted by the State are sufficiently reliable to support the truth of the proposition that the relative risk of suicide and suicide ideation is higher for women who abort *899their pregnancies compared to women who give birth or have not become pregnant. It also is worth noting that Planned Parenthood does not challenge the disclosure that “[depression and related psychological distress” is a “known medical risk[ ] of the [abortion] procedure.” S.D.C.L. § 34-23A-10.1(l)(e)(i); see also Gonzales v. Carhart, 550 U.S. 124, 159, 127 S.Ct. 1610, 167 L.Ed.2d 480 (2007) (noting that “[s]evere depression and loss of esteem can follow” an abortion). As a matter of common sense, the onset of depression and psychological distress also would increase one’s risk of suicide and suicide ideation. See, e.g., Ottar Bjerkeset et al., Gender Differences in the Association of Mixed Anxiety and Depression with Suicide, 192 Brit. J. Psychiatry 474, 474 (2008) (“Depression is thought to be the most important antecedent of suicide.... ”). Thus, there appears to be little dispute about the truthfulness of the required disclosure.

Finally, Planned Parenthood contends that the suicide advisory is not truthful because an increased risk of suicide after abortion is not “known” as required by the statute. See S.D.C.L. § 34-23A-10.1(l)(e) (requiring disclosure of “[a]ll known medical risks of the procedure”); Rounds, 653 F.3d at 673 (“ ‘[K]nown’ means generally recognized, proved, or familiar to all.”). Once again, however, this contention is premised on Planned Parenthood’s argument that the term “increased risk” implies a causal link that is not generally “known.” Because the statute does not require the disclosure of any causal link, Planned Parenthood’s argument on this point is misdirected. The record indicates that the disclosure actually required — that the relative risk of suicide and suicide ideation is higher for women who abort compared to women in other relevant groups — is generally “known.” For example, the ninety-one-page APA Report, on which Planned Parenthood relies extensively, was commissioned for the sole purpose of analyzing that “known” risk in more detail. See APA Report at 5.

As a result, we hold that the disclosure facially mandated by the suicide advisory is truthful.

V.

Despite the extensive evidence in the record of an “increased risk” of suicide, Planned Parenthood contends that disclosure of the increased risk would be misleading and irrelevant to a patient seeking an abortion, see Rounds, 530 F.3d at 735, because some authorities have indicated that there is no direct causal link. In particular, Planned Parenthood argues that it is more plausible that certain underlying factors, such as pre-existing mental health problems, predispose some women both to have unwanted pregnancies and to have suicidal tendencies, resulting in a misleading correlation between abortion and suicide that has no direct causal component. Under this view, the required disclosure would be misleading or irrelevant to the decision to have an abortion because the patient’s decision would not alter the underlying factors that actually cause the observed increased risk of suicide.

As an initial matter, the standard medical practice, as reflected in the record, is to recognize a strongly correlated adverse outcome as a “risk” while further studies are conducted to clarify whether various underlying factors play causal roles. See, e.g., Coleman Decl. ¶¶ 9-11, Jul. 6, 2006. In contravention of that standard practice, Planned Parenthood argues that the mere existence of underlying factors proscribes the disclosure of suicide as a risk related to abortion. However, there is no constitutional requirement to invert the traditional understanding of “risk” by requiring, where abortion is involved, that conclusive understanding of causation be obtained first. Indeed, the Supreme *900Court “has given state and federal legislatures wide discretion to pass legislation in areas where there is medical and scientific uncertainty,” and “[m]edieal uncertainty does not foreclose the exercise of legislative power in the abortion context any more than it does in other contexts.” Gonzales, 550 U.S. at 163-64, 127 S.Ct. 1610. In particular, “a requirement that a doctor give a woman certain information as part of obtaining her consent to an abortion is, for constitutional purposes, no different from a requirement that a doctor give certain specific information about any medical procedure.” Casey, 505 U.S. at 884, 112 S.Ct. 2791. There is no basis in the “non-misleading” and “relevant” requirements of Casey for imposing a new, stricter definition of medical risk — a standard that requires certainty of causation— simply because the medical procedure at issue is abortion.

Thus, the truthful disclosure regarding increased risk cannot be unconstitutionally misleading or irrelevant simply because of some degree of “medical and scientific uncertainty,” Gonzales, 550 U.S. at 163, 127 S.Ct. 1610, as to whether abortion plays a causal role in the observed correlation between abortion and suicide. Instead, Planned Parenthood would have to show that any “medical and scientific uncertainty” has been resolved into a certainty against a causal role for abortion. In other words, in order to render the suicide advisory unconstitutionally misleading or irrelevant, Planned Parenthood would have to show that abortion has been ruled out, to a degree of scientifically accepted certainty, as a statistically significant causal factor in post-abortion suicides. An examination of Planned Parenthood’s evidence reveals that it has not met this burden.

First, Planned Parenthood points out that the label approved by the Food and Drug Administration (“FDA”) for the abortion-inducing drug Mifeprex (mifepristone, also known as RU-486) does not list suicide or suicide ideation as a risk of using the drug, despite FDA labeling regulations requiring the listing of, inter alia, all “clinically significant adverse reactions” and “other potential safety hazards.” See 21 C.F.R. § 201.57(c)(6)(i). However, an FDA-approved label does not represent the definitive or exclusive list of risks associated with a drug. The record before us does not show whether any evidence of the link between abortion and suicide was submitted to the FDA, nor does it provide details of the FDA’s analysis, if any, of the link. Thus, the FDA-approved label for Mifeprex yields no information as to whether abortion has been ruled out as a statistically significant causal factor in post-abortion suicides.

Second, Planned Parenthood argues, and the district court found, that the American College of Obstetricians and Gynecologists (“ACOG”), a well-known professional medical organization, “rejects any suggestion that increased risk of suicide and suicide ideation are known risks of abortion.” See Rounds, 650 F.Supp.2d at 983. Unfortunately, there was no evidence from ACOG in the record for the district court to consider. The only evidence in the record pertaining to ACOG’s position is a second-hand reference in a 2005 report by the State’s expert, Dr. Elizabeth M. Shadigian, that quoted two sentences from a single ACOG Practice Bulletin: “Long-term risks sometimes attributed to surgical abortion include potential effects on ... psychological sequelae. However, the medical literature, when carefully evaluated, clearly demonstrates no significant negative impact on any of these factors with surgical abortion.” Elizabeth M. Shadigian, Report to the S.D. Task Force to Study Abortion 4, Sept. 21, 2005, ECF No. 177-4 (hereinafter “Shadigian Report”); see also Ex. O, Shadigian Dep. 137-38, ECF No. 147-15 (quoting the recitation of those lines in the Shadigian Re*901port). Dr. Shadigian further reported her opinion that ACOG’s statement was erroneous and that “ACOG seems to claim that they have adequately evaluated the medical literature, but they do not consider our study or the many other studies we evaluated.” Shadigian Report at 5. There is no other evidence in the record as to what “medical literature” ACOG considered, in what fashion it was “carefully evaluated,” whether suicide was one of the “psychological sequelae” considered, whether ACOG’s analysis received any independent peer review, or indeed whether a “Practice Bulletin” purports to be grounded in any sort of reliable scientific method at all. The two unsupported sentences from an ACOG Practice Bulletin lend no credence to the argument that abortion has been ruled out as a statistically significant causal factor in post-abortion suicides.

Third, Planned Parenthood cites the previously mentioned APA Report. The six-person Task Force on Mental Health and Abortion that authored the APA Report reviewed “50 papers published in peer-reviewed journals between 1990 and 2007 that analyzed empirical data of a quantitative nature on psychological experiences associated with induced abortion, compared to an alternative.” APA Report at 64. For some of the studies that found increased mental health risks associated with abortion, the APA Report identifies perceived methodological deficiencies, including an inability to limit the comparison group to women who carried unplanned or unwanted pregnancies to term. See id. at 68. Based on one study that attempted to account for that variable, the report states that “the best scientific evidence indicates that the relative risk of mental health problems among adult women who have an unplanned pregnancy is no greater if they have an elective first-trimester abortion than if they deliver that pregnancy.” Id. (emphases in original). In the very same sentence, however, the report states that the published literature could not provide “unequivocal evidence regarding the relative mental health risks associated with abortion per se compared to its alternatives (childbirth of an unplanned pregnancy).” Id.

The State and Intervenors argue that the APA Report is deficient in several respects. While the APA Report alleges methodological flaws in all of the studies that found a strong link between abortion and adverse mental health outcomes, it does not systematically list or analyze those flaws for each study considered. Instead, the report uses a handful of studies as illustrative examples. The State and Intervenors contend that this lack of rigor allowed the APA Report to analyze studies that found abortion to be “a benign experience for most women” less stringently than studies that found abortion to cause adverse effects. Coleman Deck ¶ 14, Sept. 16, 2008, ECF No. 290-3. For example, while the APA Report suggests that the studies showing increased risk did not compare women receiving abortions to women who carried unplanned pregnancies to term, at least three studies purportedly considered by the task force did use such a control group, and each of those studies still “definitively indicated that abortion was associated with more mental health problems.” Id. at ¶ 19. The APA Report also does not acknowledge that some of the studies showing increased risk did statistically control for other potential causal factors such as history of depression, anxiety, suicide ideation, childhood sexual abuse, physical abuse, child neuroticism, and low self-esteem. Id. at ¶ 15(c).5 As another example, although a high rate of attrition (i.e., the loss of subjects from a *902long-term study before the study is complete) is typically regarded as a methodological weakness, the APA Report downplays the significance of attrition, possibly because “the studies with the highest attrition rates ... are also the ones that provide little evidence of negative effects” of abortion. Id. at ¶ 15(d). A number of published authors in the field contacted the APA to point out these problems and ask that the APA Report be retracted. Id. at ¶¶ 28-29.

At a minimum, it appears that many published authors in the field do not accept the opinion of the APA’s six-person task force that the “best evidence” suggests that there is no real significance to the link between abortion and suicide. Even if one accepts the findings in the APA Report at face value, however, the crux of the matter is that while the APA Report states that the evidence available at the time of its review is not “sufficient to support the claim that an observed association between abortion history and mental health was caused by the abortion,” id. at 6 (emphasis added), it also concludes that the published literature is inconclusive and more research is needed “to disentangle confounding factors and establish relative risks of abortion compared to its alternatives,” id. at 72; see also id. at 68 (admitting that the published literature could not provide “unequivocal evidence regarding the relative mental health risks associated with abortion per se compared to its alternatives (childbirth of an unplanned pregnancy)”). In other words, while the APA Report finds that studies to date have not established with certainty that abortion is a causal factor in post-abortion suicide, it also acknowledges that abortion has not been ruled out as a causal factor and that currently available studies are inadequate for that purpose.6 Thus, the APA Report provides no support for the proposition that abortion has been ruled out as a statistically significant causal factor in post-abortion suicides.

Finally, the dissent relies on six recent publications submitted to this Court by Planned Parenthood as a supplement to the district court record. While the dissent suggests that these more recent publications have eliminated any uncertainty about the causal role of abortion in the increased risk of suicide, post at 909-10, the publications add little of value to the record. As an initial matter, three of the publications7 are not new analyses of data, *903but rather reviews or surveys of existing studies. As with the APA Report, it is difficult to identify a solid objective basis for the criteria employed in these reviews to identify the “best” studies and discount the others, and in the fine print they sometimes remain equivocal about the role of abortion as a causal factor. See, e.g., Robinson, supra, at 277 (“For women who have more significant [psychological] problems, the causal contribution of the abortion is not clear; a wide range of factors, both internal and external, affect women’s responses — and interact in complex ways.”).

The three remaining supplemental publications actually provide new analysis, but each suffers from apparent weaknesses. One of the publications, Julia R. Steinberg et al., Does the Outcome of a First Pregnancy Predict Depression, Suicidal Ideation, or Lower Self-Esteem? Data from the National Comorbidity Survey, 81 Am. J. Orthopsychiatry 193 (2011) (“Steinberg I”), compared self-reported mental health problems for women who carried their first pregnancy to term with women who aborted their first pregnancy, see id. at 194, while attempting to control for prepregnancy mental health, experience of sexual violence, and age at first pregnancy, see id. at 197. Mental health problems were classified as pre- or post-delivery or abortion of the first pregnancy. Id. at 195. Thus, if a woman delivered a first pregnancy, aborted a subsequent pregnancy, and suffered an adverse mental health outcome after the abortion, her adverse outcome nevertheless was classified by Steinberg I as belonging to the “delivery” comparison group, rather than the “abortion” comparison group. According to data from the Guttmacher Institute, however, approximately 45 to 47 percent of women obtaining their first abortion have previously carried at least one pregnancy to term. See Jones et al., Repeat Abortion in the United States, Guttmacher Institute, 18 (Nov. 2006), http://www. guttmacher.org/pubs/2006/ll/21/or29.pdf (listing number of prior births for women having a first abortion, based on two sets of data collected by the Guttmacher Institute at abortion provider locations). As a result, Steinberg I almost certainly shifts the outcomes for a significant number of women who aborted their second or subsequent pregnancies from the “abortion” comparison group to the “delivery” comparison group, rendering its comparison of mental health outcomes unreliable.

Another study, Julia R. Steinberg & Lawrence B. Finer, Examining the Association of Abortion History and Current Mental Health: A Reanalysis of the National Comorbidity Survey Using a Common-Riskr-Factors Model, 72 Soc. Sci. & Med. 72 (2011) (“Steinberg II”), compared self-reported mental health problems for women who had been pregnant but never aborted with those women who had aborted one or more pregnancies, while attempting to control for pre-pregnancy mental health, experience of violence, and age at first pregnancy. See id. at 77. However, Steinberg II identified post-abortion or post-delivery mental health outcomes only as those problems the participants reported they were experiencing at the time of the survey, rather than considering mental health problems that participants reported as occurring at any time after an abortion or delivery. See id. at 76-77. Therefore, it appears that Stein-berg II addresses only an arbitrarily limit*904ed window of the women’s mental health histories.

Finally, Trine Munk-Olsen et ah, Induced First-Trimester Abortion and Risk of Mental Disorder; 364 New Eng. J. Med. 332 (2011), extracted data from the Danish Civil Registration System and the Danish Psychiatric Central Register. The study considered data only for women who had no history of inpatient treatment for mental illness in their lifetime prior to the nine-month period preceding either a first abortion or a first delivery of a pregnancy. See id. at 334. For that group of women, the study analyzed data regarding inpatient or outpatient mental health contacts during the nine-month period preceding abortion or delivery and the twelve-month period following abortion or delivery. See id. While the study observed a significantly higher number of psychiatric visits after abortion as compared to after delivery, see id. at 335, it “found no significant increase in the incidence rate of psychiatric contact in the 12 months after an induced first-trimester abortion as compared with the 9-month period before the abortion” and concluded that the higher incidence in post-abortive women compared to post-delivery women likely was due solely to higher pre-existing levels of psychiatric problems for the women who sought abortions, see id. at 336. This conclusion apparently begs the question, however, by assuming that any mental distress occurring in the nine-month period prior to an abortion procedure was completely unrelated to the abortion. It seems just as plausible to assume, particularly in a population selected for having no adverse mental health history prior to that time period, that for at least some of the women, psychological distress in that time period arose in part because they had decided, or already were facing pressure from others, to undergo the abortion. Under this interpretation, the study actually tends to confirm the legislative finding that women who seek abortions are “often under stress and pressures from circumstances and from other persons, and that there exists a need for special protection of the rights of such pregnant women.” S.D.C.L. § 34-23A-1.5.

We acknowledge that these studies, like the studies relied upon by the State and Intervenors, have strengths as well as weaknesses. Like all studies on the topic, they must make use of imperfect data that typically was collected for entirely different purposes, and they must attempt to glean some insight through the application of sophisticated statistical techniques and informed assumptions. While the studies all agree that the relative risk of suicide is higher among women who abort compared to women who give birth or do not become pregnant, they diverge as to the extent to which other underlying factors account for that link. We express no opinion as to whether some of the studies are more reliable than others; instead, we hold only that the state legislature, rather than a federal court, is in the best position to weigh the divergent results and come to a conclusion about the best way to protect its populace. So long as the means chosen by the state does not impose an unconstitutional burden on women seeking abortions or their physicians, we have no basis to interfere.

In summary, although the record reflects “medical and scientific uncertainty,” Gonzales, 550 U.S. at 163, 127 S.Ct. 1610, as to whether abortion itself is a causal factor in the observed correlation between abortion and suicide, there is nothing in the record to suggest that abortion as a cause per se has been ruled out with certainty. As a result, the disclosure of the observed correlation as an “increased risk” is not unconstitutionally misleading or irrelevant under Casey and Gonzales. Indeed, physicians who provide abortions *905should be capable of reviewing the research in the field, understanding the difference between relative risk and proof of causation, and explaining it correctly to their patients. Cf. Rounds, 530 F.3d at 736 (holding that the subject matter of the biological disclosure “should be clear in context to a physician”).8 In the end, “[t]he point of informed consent laws is to allow the patient to evaluate her condition and render her best decision under difficult circumstances. Denying her up to date medical information is more of an abuse to her ability to decide than providing the information.” Lakey, 667 F.3d at 579.

Accordingly, we hold that the suicide advisory is non-misleading and relevant to the patient’s decision to have an abortion.

VI.

In conclusion, we hold that the requirements of S.D.C.L. § 34 — 23A—10.1 (1)(e)(ii) are satisfied by a disclosure that the relative risk of suicide and suicide ideation is higher for women who abort compared to women in other relevant groups, as described in the relevant medical research. The statute does not require the physician to disclose that a causal link between abortion and suicide has been proved. The disclosure is truthful, as evidenced by a multitude of studies published in peer-reviewed medical journals that found an increased risk of suicide for women who had received abortions compared to women who gave birth, miscarried, or never became pregnant. Various studies found this correlation to hold even when controlling for the effects of other potential causal factors for suicide, including pre-existing depression, anxiety, suicide ideation, childhood sexual abuse, physical abuse, child neuroticism, and low self-esteem.

Moreover, the suicide advisory is non-misleading and relevant to the patient’s decision to have an abortion, as required by Casey. It is a typical medical practice to inform patients of statistically significant risks that have been associated with a procedure through medical research, even if causation has not been proved definitively.9 While Planned Parenthood points to uncertainty as to whether abortion itself is a causal factor in the observed correlation to suicide, as opposed to other underlying factors that tend to be associated independently with both abortion and suicide, the Supreme Court “has given state and federal legislatures wide *906discretion to pass legislation in areas where there is medical and scientific uncertainty,” including “in the abortion context.” Gonzales, 550 U.S. at 163-64, 127 S.Ct. 1610. Thus, a truthful disclosure cannot be unconstitutionally misleading or irrelevant simply because some degree of medical and scientific uncertainty persists. To be sure, informed consent requirements “must be calculated to inform [a] woman’s free choice, not hinder it,” Casey, 505 U.S. at 877, 112 S.Ct. 2791, but there is no unconstitutional hindrance of the woman’s choice where, as here, the State merely is using “its regulatory authority to require a physician to provide truthful, non-misleading information relevant to a patient’s decision to have an abortion, even if that information might also encourage the patient to choose childbirth over abortion,” Rounds, 530 F.3d at 735.

On its face, the suicide advisory presents neither an undue burden on abortion rights nor a violation of physicians’ free speech rights. Accordingly, we reverse the district court’s grant of summary judgment to Planned Parenthood with respect to S.D.C.L. § 34-23A-10.1 (l)(e)(ii), direct the entry of summary judgment for the State as to that provision, and vacate the permanent injunction against the enforcement of that provision.

LOKEN, Circuit Judge,

concurring.

Though I agree with the dissent that the plain language of S.D.C.L. § 34-23A-10.1(l)(e)(ii) — “known medical risks ... to which the pregnant woman would be subjected” — strongly suggest legislative intent to require that a physician make an untruthful, misleading causation disclosure, the first two sentences of Part VI of the court’s opinion require only a disclosure as to relative risk that the physician can adapt to fit his or her professional opinion of the conflicting medical research on this contentious subject. With the facial constitutionality of the statute limited in this fashion, controlling Supreme Court precedent requires that I concur. See Gonzales v. Carhart, 550 U.S. 124, 163-68, 127 S.Ct. 1610, 167 L.Ed.2d 480 (2007).

COLLOTON, Circuit Judge,

concurring in part and concurring in the judgment.

I concur in Parts I through III of the court’s opinion, except that I find it unnecessary to consider the meaning of the hypothetical phrase “a description of all known risks of the procedure ... to which the pregnant woman would be subjected.” Ante, at 896 & n. 3. The most natural reading of S.D.C.L. § 34-23A-10.1 is that it requires the physician to present “a description of all known medical risks of the procedure,” including “[ijncreased risk of suicide ideation and suicide.” Ante, at 896. For the reasons set forth in Part III, the statutory language thus calls for a description of the relative risks of suicide ideation and suicide that are discussed in the peer-reviewed literature. See ante, at 894-96, 898. The phrase “to which the pregnant woman would be subjected” modifies “statistically significant risk factors.” The district court’s order striking “statistically significant risk factors” did not rewrite the statute to cause the phrase that follows the stricken text to modify antecedent text that was not previously modified.

I concur in Part IV of the court’s opinion concerning why the required disclosure is truthful. I also concur in the portion of Part V that explains why the record before the district court did not establish that the disclosure is misleading. This court took the unusual step of permitting the appellees to supplement the record on appeal, after the completion of briefing, with a 476-page supplemental appendix that includes several studies that were not presented to the district court. Without attempting to engage in a social science *907critique of these studies in the first instance, it is sufficient to observe that the conclusions of these studies do not, on their face, eliminate the medical and scientific uncertainty concerning the relationship between abortion and suicide ideation or suicide. See ante, at 904. I thus concur that the State was permitted to require a description of the relative risks as reflected in the peer-reviewed literature, with the physician free to augment that description based on his or her professional judgment. Ante, at 905 & n. 8.

For these reasons, I concur in the judgment.

MURPHY, Circuit Judge,

with whom WOLLMAN, BYE, and MELLOY, Circuit Judges, join, dissenting.

The record before the district court supported its conclusions that South Dakota’s 2005 suicide advisory is unconstitutional because it will not inform the free choice of a woman and is not consistent with the medical evidence. These conclusions have only been strengthened by the medical evidence received since then. The governing rule of law is that laid down by the Supreme Court in Planned Parenthood of Southeastern Pennsylvania v. Casey, 505 U.S. 833, 112 S.Ct. 2791, 120 L.Ed.2d 674 (1992), which prohibits a state from requiring an advisory which is not “calculated to inform the woman’s free choice” but “hinder[s] it.” Id. at 877, 112 S.Ct. 2791. Gonzales v. Carhart, 550 U.S. 124, 127 S.Ct. 1610, 167 L.Ed.2d 480 (2007), on which the majority relies, did not address that standard.

The most reliable evidence in the record shows that abortion does not have a causal relationship to the risk of suicide and that South Dakota’s mandated advisory is not truthful, but actually misleading. In Casey, the Court recognized both a woman’s right “to decide to terminate a pregnancy free of undue interference by the State” and the state’s “legitimate goal of ... ensuring a decision that is mature and informed” in order to “facilitate[ ] the wise exercise of that right.” 505 U.S. at 883, 887, 112 S.Ct. 2791. Focus on these parallel goals in Casey shows how carefully the Court considered the interests of both the woman and the state in that decision.

In order to be constitutional an informed consent requirement must be truthful, non misleading, and relevant. See Casey 505 U.S. at 882-83, 112 S.Ct. 2791; see also Planned Parenthood Minn., N.D., S.D. v. Rounds, 530 F.3d 724, 735 (8th Cir.2008) (en banc). Requiring physicians to provide their patients with information that does not meet this standard violates the physicians’ First Amendment right against compelled speech. Casey, 505 U.S. at 884, 112 S.Ct. 2791; see also Rounds, 530 F.3d at 734-35.

The content of the 2005 suicide advisory raises constitutional problems which the prior version of the South Dakota statute did not. The previous provision required a physician to advise a patient about the “particular medical risks associated with the particular abortion procedure to be employed, including when medically accurate, the risks of infection, hemorrhage, danger to subsequent pregnancies, and infertility.” S.D.C.L. § 34-23A-10.1(1)(b) (2003) (emphasis added). In contrast, the statute before the court requires doctors to tell a pregnant woman that a greater likelihood of suicide and suicide ideation is a “known medical risk [ ]” to which she “would be subjected” by having an abortion. S.D.C.L. § 34-23A-10.1(l)(e) (2005) (emphasis added).

The record clearly demonstrates, however, that suicide is not a known medical risk of abortion and that suicide is caused instead by factors preexisting an abortion such as a history of mental illness, domestic violence, and young age at the time of *908pregnancy. See, e.g., Julia R. Steinberg, et al., Does the Outcome of a First Pregnancy Predict Depression, Suicidal Ideation, or Lower Self-Esteem? Data from the National Comorbidity Survey, 81 Am. J. Orthopsychiatry 193 (2011); Gail Erlick Robinson, et al., Is There an “Abortion Trauma Syndrome?” Critiquing the Evidence, 17 Harv. Rev. Psychiatry 268 (2009).

As can be seen, the prior version of the South Dakota law did not carry the fatal flaw embodied in the statute now being considered. The wording of the statute under consideration conveys a causal relationship between abortion and the risk of suicide “to which the pregnant woman would be subjected.” The phrase to subject someone to something means “to cause to undergo or submit to.” Webster’s Third New Int’l Dictionary 2275 (2002). In contrast, the wording in the prior state legislation spoke of the “risks associated with ... abortion.” An association is defined as “the relationship of the occurrence of two events, without evidence that the event being investigated actually causes the second condition.” Taber’s Cyclopaedic Med. Dictionary 201 (21st ed. 2009). Legislative findings show that the statutory drafters intended that the advisory under review convey causality, for they stated that women must be informed that “procedures terminating the life of an unborn child impose risks to the life and health of the pregnant woman.” S.D.C.L. § 34-23A-1.4 (emphasis added); Webster’s Third New Int’l Dictionary 1136 (2002) (defining “impose” as “to cause to be burdened”).10

The majority concedes that there is no proof in the medical literature that abortion causes suicide, ante at 895-96, and it recognizes that an advisory telling a woman that abortion causes an increased risk of suicide would be untruthful. Ante at 897-98. It seeks to avoid the constitutional problem created by the current statutory text by suggesting that the legislature’s amendment substituting subjected to for “associated with” should not be understood to mean causality since nearly all of the words in the advisory were changed. The new language is explained as merely informing women that their decision to have an abortion would “cause[ ] [them] to become a member of a group” with a statistically higher rate of suicide. Ante at 896. That is not what the plain language of the statute says, however, and the medical evidence shows that women sharing certain factors may have a higher rate of suicide but not that abortion causes suicide.

The evidence considered by the district court shows that an advisory informing women that abortion causes them to be more likely to commit suicide is untruthful and misleading. That record made clear that abortion does not cause a “known” risk of suicide or suicide ideation. The record included volumes of deposition testimony, published medical research, and legislative reports supporting the district court’s conclusion that the suicide advisory is unconstitutional.

One of the significant reports in the record was the American Psychological Association’s (APA) review of the medical *909literature. That review showed only an association between women who have an abortion and woman who commit suicide. The APA’s review concluded that “the best scientific evidence indicates that the relative risk of mental health problems among adult women who have an unplanned pregnancy is no greater if they have an elective first-trimester abortion than if they deliver that pregnancy.” Brenda Major, et al., American Psychological Association, Report of the APA Task Force on Mental Health and Abortion 68 (2008) (APA Report).

There was also evidence from the “most recent edition of medical opinions” by the American College of Obstetricians and Gynecologists (ACOG) showing that the ACOG shared the APA’s interpretation of the medical literature and informed its members that abortion does not affect women’s subsequent mental health. The record included evidence that the label for the abortion inducing drug mifepristone was never revised to include the risk of suicide or suicide ideation. That was relevant in light of the Food and Drug Administration requirement that drug labeling must “be revised to include a warning as soon as there is reasonable evidence of an association of a serious hazard with a drug; a causal relationship need not have been proved.” 21 C.F.R. § 201.80(e).

In addition the district court was made aware of the fact that the author of two of the studies, on which the state and intervenors rely, has explained that his findings did not “support the hypothesis that abortion itself causes suicide.” Mika Gissler, et al., Letter to the Editor: Pregnancy-Related, Violent Deaths, 27 Scand. J. Pub. Health 54, 55 (1999). Gissler concluded that “[a] more likely explanation is that the excess risk may be due to causes related both to induced abortion and violent death.” Id.

The record included other criticisms of studies presented by the state and the intervenors which had used comparator groups irrelevant to a pregnant woman’s decision to have an abortion. Because pregnant women can no longer choose not to become pregnant, providing them information about the relative risks of suicide for women after abortion compared with women with no pregnancy does nothing to inform their decision on whether to have an elective abortion. See, e.g., Mika Gissler, et al., Injury Deaths, Suicides and Homicides Associated with Pregnancy, Finland 1987-2000, 15 European J. Pub. Health 459, 460 (2005) (comparing women electing abortion with women who are not pregnant); David M. Fergusson, et al., Abortion in Young Women and Subsequent Mental History, 47 J. Child Psychol. & Psychiatry 16, 17 (2006) (same); see also APA Report at 53-54, 71 (discussing this methodological problem).

Since the district court enjoined the suicide advisory and a panel of this court affirmed that decision, the United Kingdom’s Royal College of Obstetricians and Gynaecologists (RCOG) has issued recommendations that women “be informed that the evidence suggests that they are no more or less likely to suffer adverse psychological sequelae whether they have an abortion or continue with the pregnancy and have the baby.” RCOG, The Care of Women Requesting Induced Abortion 45 (Nov. 2011). The United Kingdom’s National Collaborating Centre for Mental Health arrived at the same conclusion in its report to the Academy of Medical Royal Colleges. Induced Abortion and Mental Health: A Systemic Review of the Mental Health Outcomes of Induced Abortion, Including Their Prevalence and Associated Factors 125 (Dec. 2011).

These conclusions are based on numerous studies which strengthen the evidence *910on which the district court relied. The studies establish that post abortion suicide rates are linked to preexisting mental illness and domestic violence, not to the decision to undergo an abortion. See, e.g., Trine Much-Olsen, et al., Induced First-Trimester Abortion and Risk of Mental Disorder, 364 New Eng. J. Med. 332, 338 (2011); Robinson, supra, at 276 (“The most well controlled studies continue to demonstrate that there is no convincing evidence that induced abortion of an unwanted pregnancy is per se a significant risk factor for psychiatric illness.”). If, as the majority points out, “the standard medical practice ... is to recognize a strongly correlated adverse outcome as a ‘risk’ while further studies are conducted to clarify whether various underlying factors play causal roles,” ante at 899 (emphasis omitted), must not research conducted by experts in the field after the district court’s decision be considered as corroboration of its findings and conclusions?

Dr. Priscilla Coleman, an expert witness produced by the state and intervenors in the district court, has recently been criticized for her study methodology and her resulting conclusions that abortion plays a causal role to increase the risk of suicide. In one study researchers used the same data and methodology Coleman had in a 2009 study discussed in one of her declarations to the district court. Guhin Deck, Exh. 87 at 13, ECF No. 290-2. The researchers found that Coleman’s results were not replicable and concluded that “structural, psychological, and sociodemographic risk factors associated with both having an abortion and having poor mental health drive a relationship between abortion and mental health.” Julia R. Stein-berg & Lawrence B. Finer, Examining the Association of Abortion History and Current Mental Health: A Reanalysis of the National Comorbidity Survey Using a Common-Risk-Factors Model, 72 Soc. Sci. & Med. 72, 81 (2011). The editor-in-chief of the Journal of Psychiatric Research subsequently concluded that Coleman’s explanation for her methodology in the 2009 study was “unpersuasive” and that the analysis “does not support [Coleman’s] assertions” that abortions “were associated with increased risk of lifetime mental disorders----” Reply to Letter to the Editor: Commentary on Abortion Studies of Steinberg and Finer (Soc. Sci. & Med. 2011; 72:72-82) and Coleman (J. Psychiatric Res. 2009; 13:770-6 & J. Psychiatric Res. 2011; 15:1133-1), 46 J. Psychiatric Res. 410, 410 (2012).

The quality of the cited studies has been recognized by leading professional associations. This research also formed the basis for the opinions of these bodies that the induced abortion of an unwanted pregnancy does not cause an increased risk of mental health problems. See, e.g., National Collaborating Centre for Mental Health, supra at 125-27. Rather than recognizing this emerging consensus based on the scientific research in the record before the district court and all the subsequently submitted evidence by the parties to this court, the majority theorizes about the nature of an advisory. In the end it arrives at a new test divorced from the standard established in Casey.

The majority posits that the lack of evidence — that the correlation between abortion and suicide is due to a causal relationship — is not fatal to the advisory because the existence of a correlation for any reason makes the advisory truthful. Pointing out that Planned Parenthood does not currently challenge the state’s depression advisory, it asserts that “as a matter of common sense” depression can be a precursor to suicide. Ante at 899. While Planned Parenthood withdrew its challenge to that section of the statute, it never conceded that “depression and related psychological *911distress” are known medical risks of abortion nor does it inform its patients of this. Resp. Pet. for Reh’g n. 8. Even a study submitted by the intervenors admits that data do not support an association between abortion and depression. David M. Fergusson, et al. A Further Meta-Analysis, Br. J. of Psychiatry, Oct. 5, 2011 available at http://bjp.rcpsych.Org/content/199/3/180/ reply#bjprepsych_eL.33839.11

The majority concedes though that if the correlation between abortion and suicide were not due to a causal relationship, then the advisory “would be misleading or irrelevant to the decision to have an abortion because the patient’s decision would not alter the underlying factors that actually cause the observed increased risk of suicide.” Ante at 899. The vast majority of researchers, however, assert that this is precisely the case. Those studies in the record show that other independent factors which co-occur with both abortion and suicide, such as prepregnancy mental health problems, domestic violence, and youth, account for the correlation between abortion and suicide risk.

To overcome this evidentiary problem a new standard for informed consent advisories is offered. Under this proposed test, so long as a causal link between abortion and suicide would be theoretically possible, an advisory is truthful, non misleading, and relevant unless Planned Parenthood can prove the absence of a causal link with “scientifically accepted certainty.” Ante at 900. In support the court turns to Gonzales, 550 U.S. at 163-67, 127 S.Ct. 1610, to rely on its discussion of medical uncertainty. Ante at 899-900, 904-05. The Court there was not considering a Casey issue about informed consent, however, and it was not evaluating the information given to an individual woman to “ensur[e] a decision that is mature and informed.” See Casey, 505 U.S. at 883, 112 S.Ct. 2791. The Court concluded only that Congress, which was fully informed of the contradicting medical opinions, could balance the need to protect the state’s interests in the “ethics of the medical profession” and “respect for dignity of human life” against the uncertain risks to women’s health resulting from the ban. Gonzales, 550 U.S. at 157, 166, 127 S.Ct. 1610 (citation omitted).

The state’s interest in this case is to promote a “wise,” “mature[,] and informed” decision by women considering abortion. Casey, 505 U.S. at 883, 887, 112 S.Ct. 2791. Here, any medical uncertainty as to whether abortion causes an increased risk of suicide undermines the advisory’s constitutionality because a woman’s ability to make a wise, mature, and informed choice is hindered by being told that the increased risk of suicide is a “known medical risk[ ]” “to which ... [she] would be subjected” by having an abortion when the weight of the medical research indicates the opposite and she is not informed of the debate. The state’s interest is thus not furthered by such an advisory.

It is significant that the South Dakota legislature and governor amended certain abortion regulations in March 2012 in order to reflect the more accepted view in the medical community that abortion does not cause mental health problems such as suicidal ideation and suicide. In the new version of the statute, which requires a physician to meet with a pregnant woman before she can schedule an abortion, the *912state legislature eliminated language mandating an assessment “to determine if any of the risk factors associated with abortion are present in her case.” S.D. House Bill 1254 § 2 ¶ 4 (amending S.D.C.L. § 34-23A-56). The state law now requires an assessment “to determine if any of the following preexisting risk factors associated with adverse psychological outcomes following an abortion are present in her case.” Id. Among the listed preexisting risk factors in South Dakota’s revision are coercion, a history of mental illness, and youth. Id. This amendment thus brings the statute in line with the existing medical evidence which shows that an increased risk of suicide is linked not to the decision to undergo an abortion, but to preexisting risk factors that coincide with abortion.

We agree that “[t]he point of informed consent laws is to allow the patient to evaluate her condition and render her best decision under difficult circumstances” and that “[d]enying her up to date medical information is more of an abuse to her ability to decide than providing the information.” See ante at 905 (quoting Tex. Med. Providers Performing Abortion Servs. v. Lakey, 667 F.3d 570, 579 (5th Cir.2012)) (first alteration in original). Yet, instead of recognizing that medical research has shown that South Dakota’s suicide advisory is untruthful, misleading, and irrelevant, the majority tries to shift the responsibility to attending, physicians to “review! ] the research in the field, understand! ] the difference between relative risk and proof of causation, and explain! ] it correctly to their patients.” Ante at 905. The statute provides only for a written transaction between doctor and patient in which explanation and clarification occur if a woman requests it, see S.D.C.L. § 34-23A-10.1 ¶¶ 2, 3, but no judicial attempt to direct the content of the conversation between a patient and her doctor can remedy the advisory’s constitutional shortcomings.

By forcing doctors to inform women that abortion subjects them to a risk which the record medical evidence refutes, the suicide advisory places an undue burden on a pregnant woman’s due process rights and violates a doctor’s First Amendment right against compelled speech. The district court’s order enjoining the suicide advisory should therefore be affirmed.