*1149OPINION

PREGERSON, Circuit Judge:

This 42 U.S.C. § 1983 case concerns the constitutionality of Los Angeles Municipal Code Section 85.02, which prohibits use of a vehicle “as living quarters either overnight, day-by-day, or otherwise.” Plaintiffs include four homeless individuals who parked their vehicles in the Venice area of Los Angeles and were cited and arrested for violating Section 85.02. Defendants are the City of Los Angeles and individual LAPD officers. Plaintiffs argue that Section 85.02 is unconstitutionally vague on its face because it provides insufficient notice of the conduct it penalizes and promotes arbitrary and discriminatory enforcement. We agree.

FACTUAL BACKGROUND

I. Section 85.02 and the Venice Homelessness Task Force

In 1983, the City of Los Angeles enacted Municipal Code Section 85.02:

USE OF STREETS AND PUBLIC PARKING LOTS FOR HABITATION.

No person shall use a vehicle parked or standing upon any City street, or upon any parking lot owned by the City of Los Angeles and under the control of the City of Los Angeles or under control of the Los Angeles County Department of Beaches and Harbors, as living quarters either overnight, day-by-day, or otherwise.

On September 23, 2010, Los Angeles officials held a “Town Hall on Homelessness” to address complaints of homeless individuals with vehicles living on local streets in Venice. Present at the meeting were a member of the City Council, the Chief of the LAPD, the Chief Deputy to the City Attorney, and the Assistant Director of the Los Angeles Bureau of Sanitation. City officials repeated throughout the meeting that their concern was not homelessness generally, but the illegal dumping of trash and human waste on city streets that was endangering public health. To address this concern, officials announced a renewed commitment to enforcing Section 85.02.

Within the week, the LAPD created the Venice Homelessness Task Force (the “Task Force”). The Task Force’s twenty-one officers were to use Section 85.02 to cite and arrest homeless people using their automobiles as “living quarters,” and were also to distribute to such people information concerning providers of shelter and other social services.

Defendant Captain Jon Peters ran the Task Force, which included Defendant Officers Randy Yoshioka, Jason Prince, and Brianna Gonzales. Task Force officers received informal, verbal training, as well as internal policy memoranda, on how to enforce Section 85.02. Supervisors instructed officers to look for vehicles containing possessions normally found in a home, such as food, bedding, clothing, medicine, and basic necessities. According to those instructions, an individual need not be sleeping or have slept in the vehicle to violate Section 85.02. Supervisors directed officers to issue a warning and to provide information concerning local shelters on the first instance of a violation, to issue a citation on the second instance, and to make an arrest on the third.

II. Enforcement of Section 85.02

Beginning in late 2010, the Task Force began enforcing Section 85.02 against homeless individuals. Four such homeless individuals are Plaintiffs in this case:1

*1150Plaintiff Steve Jaeobs-Elstein ran his own legal temp company for almost ten years before losing his business and his home in the economic downturn of 2007. He subsequently suffered severe anxiety and depression. He was able to keep his car, a small SUV, and pay for insurance, maintenance, and gas with the $200 he collects each month from General Relief. He kept his few possessions—mainly two computers and some clothes—in his car because he could not afford storage fees.

When Jaeobs-Elstein first became homeless, he slept in his car. In mid-2009, an LAPD officer approached Jaeobs-Elstein while parked on a city street, warning him that if he slept in his vehicle at night on public streets he would be arrested. At the time, Jaeobs-Elstein was unaware that such conduct was unlawful. He then looked up Section 85.02 on the Internet and, based on what he read and what the officer told him, understood Section 85.02 to mean that he could not sleep in his car on a public street in Los Ange-les. He began sleeping at motels and on other private property, and soon obtained permission from a Methodist Church in Venice to sleep in his car while it was parked in the church parking lot, provided he leave the lot by 8:00 a.m. each day. He also registered with the People Assisting The Homeless’s “Venice Vehicles to Homes” program, secured a spot on the housing wait lists maintained by the Department of Mental Health and the Los Angeles Housing Authority, and was approved for a Section 8 housing voucher through the Department of Housing and Urban Development.

On the morning of September 13, 2010, Jaeobs-Elstein was waiting in his car on a public street for the First Baptist Church of Venice to open so that he could volunteer to serve at the food distribution program, and also receive a meal. That morning, Defendant Officer Gonzales and her partner ordered Jaeobs-Elstein out of his car, searched his car, and cited him for violating Section 85.02. The officers provided him no shelter or social services information.

A few weeks later, Jaeobs-Elstein was again waiting in his car on a public street for First Baptist to open when Officer Gonzales banged on the driver’s side window and told Jaeobs-Elstein it was illegal to live in his vehicle. Two weeks later, Gonzales and her partner again spotted Jaeobs-Elstein, this time when he was parked legally in the First Baptist parking lot, and yelled at him from across the street that the next time they saw him they would take him to jail.

On the morning of October 31, 2010, Jaeobs-Elstein was exiting his car when Officer Gonzales and her partner detained, handcuffed, and arrested Jaeobs-Elstein for violating Section 85.02. The car contained personal belongings, such as boxes and computer equipment, as well as plastic bottles of urine. Jaeobs-Elstein was in custody for about seven hours before being released, after which he borrowed money to get his car out of impoundment. He had no criminal record before this arrest.

*1151On January 30, 2011, Defendant Officer Yoshioka and his partner cited Jacobs-Elstein again for violating Section 85.02, this time while Jacobs-Elstein was sitting in his car, talking on his cell phone. Jacobs-Elstein had dog food in the car. He told Officer Yoshioka the dog food was from a friend whose dog he would later take to the park. The car also contained salad boxes, water bottles, a portable radio, and bags of clothes. Jacobs-Elstein showed Officer Yoshioka proof that he resided on private property, and thus was not sleeping in his vehicle. Officer Yoshio-ka informed him that he need not sleep in his car to violate Section 85.02.

During this last incident, Officer Yoshio-ka’s partner gave Jacobs-Elstein a “Local Resources Information” pamphlet. This was the first time he was offered any such information. The flyer claimed to provide guidance on how to comply with Section 85.02. Yet Jacobs-Elstein soon discovered that this information was not helpful to him. It provided information only on RV parks, where Jacobs-Elstein could not park his car, and shelters, where he could not keep his belongings during the day.

Plaintiff Chris Taylor sells his artwork at a booth on Venice Beach, where he works every day. In October 2010, Officer Yoshioka issued a warning to Taylor for sleeping in his small two-door car through the night, in violation of Section 85.02. He then began sleeping on the sidewalk, which is legal. Starting December 1, 2010, Taylor began sleeping at Winter Shelter in Culver City. He rented a storage facility to get his excess property out of the ear, though he kept his sleeping bag with him in case he missed the bus to the shelter and had to sleep on the streets.

On the morning of December 18, 2010, Officer Yoshioka and his partner arrested Taylor for violating Section 85.02 and had his car impounded. At the time he was arrested, Taylor was sitting in his car to get out of the rain. The vehicle contained one tin of food, clothing, and a bottle of urine. Taylor informed the officers that he slept at Winter Shelter and not in his car, and that he had an identification card issued by Winter Shelter to prove it. He was arrested nonetheless.

Plaintiff Patricia Warivonchik has lived in Venice for thirty-four years. She is epileptic, and after suffering a significant head injury, is unable to work full time. Because she could no longer afford to pay rent in Venice, but did not want to leave the area, she began living in her RV. Since becoming homeless, Warivonchik has supported herself with part-time jobs and by selling ceramic artwork. She is also a member of a church in Santa Monica where she legally parks her RV at night.

On November 13, 2010, Warivonchik was driving her RV through Venice—taking her artwork to a local fair—when she was pulled over by Officer Yoshioka and his partner for failing to turn off her left blinker. She was not cited for the blinker, but was given a written warning for violating Section 85.02 and told that she would be arrested if ever seen again in Venice with her RV.

Plaintiff William Cagle has been a resident of Venice since 1979. He suffers from congestive heart failure, which causes fluid to build up in his legs, preventing him from walking even short distances. His sole source of income is Social Security, which is not enough to pay both for rent and for the medicine he needs that is not covered by his insurance. Cagle became homeless in 1993, but was able to keep his small van.

In the early mornings of October 17, 2010, and November 22, 2010, Officer Yo-shioka and his partner cited and arrested Cagle for violating Section 85.02. Among the items found in Cagle’s van were clothing, bedding, boxed food, bottles of medi*1152cine, and a portable radio. Cagle explained to the officers that he was not sleeping in his vehicle. Officer Yoshioka’s partner responded that sleeping is not the only criteria for violating Section 85.02.

PROCEDURAL HISTORY

I. The Complaint

In their First Amended Complaint, Plaintiffs challenged Section 85.02 under the Fourth, Fifth, and Fourteenth Amendments, various sections of the California Constitution, and several state and federal statutes. Although Plaintiffs alleged that enforcement of Section 85.02 “violates due process,” they did not specifically allege that the statute is unconstitutionally vague.

II. Discovery

The parties proceeded to discovery. Plaintiffs filed a discovery request for “[a]ny and all documents regarding the incident(s) described in the Complaint.” On August 22, 2011—eight days before the discovery cut-off date—Defendants filed their tenth response to Plaintiffs’ discovery request. In their response, Defendants for the first time produced copies of internal memoranda instructing officers on how to enforce Section 85.02.

In one memo from 2008, officers were told that any arrest “report must describe in detail observations ... that establish one of the following—(i) overnight occupancy for more than one night or (ii) day-by-day occupancy of three or more days.” The arrest reports for Plaintiffs Jacobs-Elstein, Taylor, and Cagle, however, contained no such observations. In another memo, from 2010, officers were told to “adhere to the ‘Four C’s’ philosophy: Commander’s Intent, Constitutional Policing, Community Perspective, and Compassion,” with no further details.

On August 26, 2011, Plaintiffs’ attorney deposed the Task Force’s lead officer, Defendant Captain Jon Peters. Plaintiffs’ attorney questioned Captain Peters extensively on whether the Task Force had been given any limiting instructions on how to enforce Section 85.02. Specifically, Plaintiffs’ attorney asked about the 2008 memo directing officers to make an arrest only after observing a suspect occupying a vehicle for more than one night or for three consecutive days, an instruction Defendant Officers had ignored. Captain Peters then stated that he disapproved of this memo because he felt it did not offer Task Force officers enough discretion, and had instead instructed officers to follow the broadly-worded' “Four C’s” policy. Plaintiffs’ attorney asked Captain Peters if he believed a person who slept at a shelter but was found in her vehicle during the day would be in violation of Section 85.02. Captain Peters responded, “I don’t believe that they would be violating the law, in my opinion.”

On August 30, 2011, Plaintiffs’ attorney deposed Defendant Officer Jason Prince. Again, Plaintiffs’ attorney repeatedly asked whether Task Force officers had been given any specific training or guidance on how to enforce Section 85.02, particularly if a suspect did not sleep in the vehicle at night. Officer Prince responded, “The totality of the circumstances is what brings us to the conclusion that they’re in violation of [Section] 85.02, not where they’re sleeping at nighttime.”

After those two depositions revealed conflicting views among the enforcing officers as to what Section 85.02 means, Plaintiffs’ attorney told Defense counsel that Plaintiffs would now be challenging the constitutionality of Section 85.02 on vagueness grounds. On September 13, 2013, Plaintiffs’ attorney emailed Defense counsel confirming that one of Plaintiffs’ “primary arguments [is] vagueness,” then mentioned three Supreme Court cases dis*1153cussing the void-for-vagueness doctrine: Papachristou v. City of Jacksonville, 405 U.S. 156, 92 S.Ct. 839, 31 L.Ed.2d 110 (1972), Kolender v. Lawson, 461 U.S. 352, 103 S.Ct. 1855, 75 L.Ed.2d 903 (1983), and City of Chicago v. Morales, 527 U.S. 41, 119 S.Ct. 1849, 144 L.Ed.2d 67 (1999).

III. Motions for Summary Judgment

On September 14, 2011, the parties filed cross-motions for summary judgment. In their motion, Plaintiffs argued that “ § 85.02 is unconstitutionally vague and criminalizes otherwise innocent behavior with insufficient notice as to what constitutes a violation of the law.... Section 85.02 is totally devoid of any standards or guidelines to limit police discretion in enforcing a vague law.”

On September 26, 2011, Plaintiffs filed their opposition to Defendants’ motion for summary judgment, again raising the argument that Section 85.02 is impermissibly vague.

That same day, Defendants filed their opposition to Plaintiffs’ motion for summary judgment. As to Plaintiffs’ vagueness challenge, Defendants first argued that “Plaintiffs’ allegations and theories of liability are confined to those found in the operative complaint,” and that Defendants were not on notice that vagueness would be at issue during summary judgment. Defendants went on, however, to defend Section 85.02 against Plaintiffs’ vagueness challenge, on the merits.

On October 3, 2011, Plaintiffs filed their reply in support of their motion for summary judgment. In it, Plaintiffs explained to the district court that it was not until eight days before the end of discovery that Defendants disclosed the LAPD’s internal memoranda describing the discretion officers had in enforcing Section 85.02. This was “significant since, when faced with a vagueness challenge to a municipal ordinance, courts are required to consider any possible limiting instructions.... ” Thus, “[Defendants can hardly complain when they only turned over key documents a week before the end of discovery.” Plaintiffs also informed the court that Plaintiffs’ attorney had told Defense counsel on August 30, 2011, that Plaintiffs would now be raising a vagueness challenge, and sent an email confirming this on September 13, 2011.

On October 28, 2011, the district court denied Plaintiffs’ motion for summary judgment and granted Defendants’ motion for summary judgment as to all claims. In a footnote, the district court held that because Plaintiffs failed to raise a vagueness challenge in their First Amended Complaint, “Defendants were not on notice that Plaintiffs would challenge the constitutionality of § 85.02 [on vagueness grounds] and such arguments are inappropriate.”

Plaintiffs timely appeal.2 We have jurisdiction under 28 U.S.C. § 1291.

STANDARD OF REVIEW

We review de novo a grant or denial of summary judgment “to determine whether, viewing the evidence in a light most favorable to the nonmoving party, there are any genuine issues of material fact and whether the district court applied the relevant substantive law.” Tzung v. State Farm Fire & Cas. Co., 873 F.2d 1338, 1339-40 (9th Cir.1989) (internal citation omitted).

*1154DISCUSSION

I. The district court abused its discretion by not addressing Plaintiffs’ vagueness claim on the merits.

The district court refused to consider the merits of Plaintiffs’ vagueness challenge because it was not expressly raised in their First Amended Complaint. That ruling was an abuse of discretion: Plaintiffs should have been granted leave to amend their First Amended Complaint to add their new claim.

Plaintiffs made their vagueness argument both in their motion for summary judgment and in their opposition to Defendants’ motion for summary judgment. Where plaintiffs “fail[] to raise [a claim] properly in their pleadings, ... [if] they raised it in their motion for summary judgment, they should [be] allowed to incorporate it by amendment under Fed.R.Civ.P. 15(b).” Jackson v. Hayakawa, 605 F.2d 1121, 1129 (9th Cir.1979). And “when issues are raised in opposition to a motion to summary judgment that are outside the scope of the complaint, ‘[t]he district court should have construed [the matter raised] as a request pursuant to rule 15(b) of the Federal Rules of Civil Procedure to amend the pleadings out of time.’ ” Apache Survival Coal. v. United States, 21 F.3d 895, 910 (9th Cir.1994) (quoting Johnson v. Mateer, 625 F.2d 240, 242 (9th Cir.1980)).

“[L]eave to amend ‘shall be freely given when justice so requires,’ Fed. R.Civ.P. 15(a), and this policy is to be applied with extreme liberality.” Morongo Band of Mission Indians v. Rose, 893 F.2d 1074, 1079 (9th Cir.1990). “Five factors are taken into account to assess the propriety of a motion for leave to amend: bad faith, undue delay, prejudice to the opposing party, futility of amendment, and whether the plaintiff has previously amended the complaint.” Johnson v. Buckley, 356 F.3d 1067, 1077 (9th Cir.2004). “The denial of a motion to amend a complaint is reviewed for abuse of discretion.” Id.

First, there is no evidence of bad faith. Second, there was no undue delay because Plaintiffs only fully understood Defendants’ enforcement policies late in the discovery period. Defendants made Plaintiffs aware of the LAPD’s 2008 and 2010 internal memoranda—describing the Task Force’s policy of enforcement—eight days before the discovery cut-off. As discussed in.Part II.B below, the vagueness analysis of a statute includes a review of any limiting interpretation adopted by the enforcement agency. These two memoranda alerted Plaintiffs that Task Force officers had either received ambiguous instructions, or had ignored the explicit directives they had been given. Once Plaintiffs received these key documents, they advanced their vagueness argument.

Third, there was no prejudice to Defendants. The district court found that Defendants were not on notice that Plaintiffs would raise a vagueness challenge at summary judgment. Yet the record shows otherwise. After finally receiving Defendants’ 2008 and 2010 internal memoranda, Plaintiffs’ attorney repeatedly asked Defendants during their depositions whether Task Force officers had any criteria to limit their enforcement of Section 85.02, especially when it came to suspects—like Plaintiffs—who did not spend the night in their vehicles. This questioning put Defendants on notice that Plaintiffs were concerned with the vagueness of Section 85.02 and the lack of limiting instructions provided by the LAPD.

Once Plaintiffs fully understood Defendants’ policy of enforcing Section 85.02, Plaintiffs confirmed that they sought to challenge Section 85.02 on vagueness grounds. Plaintiffs’ attorney told Defense counsel weeks before the parties filed *1155cross-motions for summary judgment that Plaintiffs would be raising a vagueness challenge, and repeated this statement by email the day before cross-motions for summary judgment were filed.

By the summary judgment stage, Defendants had ample notice of Plaintiffs’ vagueness challenge, and the issue did not require further discovery. Both parties fully argued the vagueness issue in their respective summary judgment briefings. Thus, any claim of surprise or prejudice by Defendants is unpersuasive. See Howey v. United States, 481 F.2d 1187, 1191 (9th Cir.1973) (finding no undue prejudice when defendant “was fully prepared to litigate” new issues raised in amended complaint).

Fourth, there is no showing that amendment would be futile. And fifth, Plaintiffs only amended their complaint once, long before they received Defendants’ internal memoranda.

The district court should have construed Plaintiffs’ vagueness argument at summary judgment as a motion to amend their First Amended Complaint. And given Defendants’ late disclosures and inability to make a credible claim of surprise or prejudice, the district court abused its discretion by not amending the First Amended Complaint to conform to the evidence and argument, and by not considering the vagueness claim on the merits.

II. Section 85.02 is unconstitutionally vague.

A statute fails under the Due Process Clause of the Fourteenth Amendment “if it is so vague and standardless that it leaves the public uncertain as to the conduct it prohibits.... ” Giaccio v. Pennsylvania, 382 U.S. 399, 402, 86 S.Ct. 518, 15 L.Ed.2d 447 (1966). A statute is vague on its face when “no standard of conduct is specified at all. As a result, men of common intelligence must necessarily guess at its meaning.” Coates v. City of Cincinnati 402 U.S. 611, 614, 91 S.Ct. 1686, 29 L.Ed.2d 214 (1971) (internal quotation marks omitted).

“Vagueness may invalidate a criminal law for either of two independent reasons. First, it may fail to provide the kind of notice that will enable ordinary people to understand what conduct it prohibits; second, it may authorize and even encourage arbitrary and discriminatory enforcement.” Morales, 527 U.S. at 56, 119 S.Ct. 1849 (citation omitted). Section 85.02 fails under both standards.

A. Section 85.02 fails to provide adequate notice of the conduct it criminalizes.

“[T]he purpose of the fair notice requirement is to enable the ordinary citizen to conform his or her conduct to the law.” Id. at 58, 119 S.Ct. 1849. A penal statute cannot require the public to speculate as to its meaning while risking life, liberty, and property in the process. See Lanzetta v. New Jersey, 306 U.S. 451, 453, 59 S.Ct. 618, 83 L.Ed. 888 (1939).

Section 85.02 offers no guidance as to what conduct it prohibits, inducing precisely this type of impermissible speculation and uncertainty. It states that no person shall use a vehicle “as living quarters either overnight, day-by-day, or otherwise.” Yet the statute does not define “living quarters,” or specify how long—or when— is “otherwise.” We know that under Defendants’ enforcement practices sleeping in a vehicle is not required to violate Section 85.02, as Jacobs-Elstein learned, nor is keeping a plethora of belongings required, as Taylor learned. But there is no way to know what is required to violate Section 85.02.

Instead, Plaintiffs are left guessing as to what behavior would subject them to citation and arrest by an officer. Is it impermissible to eat food in a vehicle? Is it *1156illegal to keep a sleeping bag? Canned food? Books? What about speaking on a cell phone? Or staying in the car to get out of the rain? These are all actions Plaintiffs were taking when arrested for violation of the ordinance, all of which are otherwise perfectly legal. And despite Plaintiffs’ repeated' attempts to comply with Section 85.02, there appears to be nothing they can do to avoid violating the statute short of discarding all of their possessions or their vehicles, or leaving Los Angeles entirely. All in all, this broad and cryptic statute criminalizes innocent behavior, making it impossible for citizens to know how to keep their conduct within the pale.

In this respect, Section 85.02 presents the same vagueness concerns as the anti-loitering ordinance held unconstitutional in Morales, 527 U.S. 41, 119 S.Ct. 1849. There, the Supreme Court found that a Chicago law prohibiting “loitering,” which it defined as “remainfing] in any one place with no apparent purpose,” lacked fair notice, as it was “difficult to imagine how any citizen ... standing in a public place with a group of people would know if he or she had an ‘apparent purpose.’ ” Id. at 56-57, 119 S.Ct. 1849.

So too here. It is difficult to imagine how anyone loading up his or her car with personal belongings, perhaps to go on a camping trip or to donate household wares to the Salvation Army, and parking briefly on a Los Angeles street, would know if he or she was violating the statute. What’s worse, even avoiding parking does not seem to be sufficient; Plaintiff Warivon-chik was not even parked—she was driving her RV through Venice when she was pulled over and issued a warning. So, under the Task Force’s expansive reading of this already amorphous statute, any vacationer who drives through Los Angeles in an RV may be violating Section 85.02. As “the [C]ity cannot conceivably have meant to criminalize each instance a citizen” uses a vehicle to store personal property, vagueness about what is covered and what is not “dooms this ordinance.” Id. at 57, 119 S.Ct. 1849.

Because Section 85.02 fails to draw a clear line between innocent and criminal conduct, it is void for vagueness.

B. Section 85.02 promotes arbitrary enforcement that targets the homeless.

A statute is also unconstitutionally vague if it encourages arbitrary or discriminatory enforcement. See Papachristou, 405 U.S. at 162, 92 S.Ct. 839. If a statute provides “no standards governing the exercise of ... discretion,” it becomes “a convenient tool for harsh and discriminatory enforcement by local prosecuting officials, against particular groups deemed to merit their displeasure.” Id. at 170, 92 S.Ct. 839 (internal quotation marks omitted).

Arbitrary and discriminatory enforcement is exactly what has occurred here. As noted, Section 85.02 is broad enough to cover any driver in Los Angeles who eats food or transports personal belongings in his or her vehicle. Yet it appears to be applied only to the homeless. The vagueness doctrine is designed specifically to prevent this type of selective enforcement, in which a “ ‘net [can] be cast at large, to enable men to be caught who are vaguely undesirable in the eyes of the police and prosecution, although not chargeable in any particular offense.’” Id. at 166, 92 S.Ct. 839 (quoting Winters v. New York, 333 U.S. 507, 540, 68 S.Ct. 665, 92 L.Ed. 840 (1948) (Frankfurter, J., dissenting)).

Section 85.02 raises the same concerns of discriminatory enforcement as the ordinance in Papachristou, 405 U.S. 156, 92 S.Ct. 839. There, the Supreme Court held that a city ordinance prohibiting “vagran*1157cy”—which was applied to “loitering,” “prowling,” and “nightwalking,” among other conduct—was unconstitutionally vague. Id. at 158, 163, 92 S.Ct. 839. The Court viewed the ordinance in its historical context as the descendant of English feudal poor laws designed to prevent the physical movement and economic ascension of the lower class. Id. at 161-62, 92 S.Ct. 839. In America, such laws had been used to “roundup ... so-called undesirea-bles,” and resulted “in a regime in which the poor and the unpopular [we]re permitted to stand on a public sidewalk ... only at the whim of any police officer.” Id. at 170, 171, 92 S.Ct. 839 (internal quotation marks omitted). The Court concluded that “the rule of law implies equality and justice in its application. Vagrancy laws ... teach that the scales of justice are so tipped that even-handed administration of the law is not possible. The rule of law, evenly applied to minorities as well as majorities, to the poor as well as the rich, is the great mucilage that holds society together.” Id. at 171, 92 S.Ct. 839.

The City argues that its enforcement goals were motivated by legitimate health and safety concerns. It notes that some of the plaintiffs were arrested while in cars with garbage, pets, and their personal belongings, and that it was unsafe for plaintiffs to occupy their cars under these circumstances. We do not question the legitimacy of these public health and safety issues, but the record plainly shows that some of the conduct plaintiffs were engaged in when arrested—eating, talking on the phone, or escaping the rain in their vehicles—mimics the everyday conduct of many Los Angeles residents. The health and safety concerns cited by the City do not excuse the basic infirmity of the ordinance: It is so vague that it fails to give notice of the conduct it actually prohibits. As shown by the City’s own documents, the different ways the ordinance was interpreted by members of the police department make it incompatible with the concept of an even-handed administration of the law to the poor and to the rich that is fundamental to a democratic society.

Defendants correctly note that they can bring clarity to an otherwise vague statute “through limiting constructions given ... by the ... enforcement agency.” Hess v. Bd. of Parole & Post-Prison Supervision, 514 F.3d 909, 914 (9th Cir.2008). Defendants point to their 2008 internal memorandum instructing officers making an arrest to first “establish one of the following—(i) overnight occupancy for more than one night or (ii) day-by-day occupancy of three or more days.” This memo is irrelevant. First, Defendant Captain Peters, who heads the Task Force, admitted that he disfavored these instructions, and instead advised his officers to adhere to the “Four C’s” philosophy, which gave Task Force officers no more guidance than the statute itself. Second, even if Task Force officers had been given the 2008 memo, they did not follow it. Officers did not observe Plaintiffs in their vehicles overnight or for three consecutive days before arresting them.

In sum, Section 85.02 has paved the way for law enforcement to target the homeless and is therefore unconstitutionally vague.

CONCLUSION

Section 85.02 provides inadequate notice of the unlawful conduct it proscribes, and opens the door to discriminatory enforcement against the homeless and.the poor. Accordingly, Section 85.02 violates the Due Process Clause of the Fourteenth Amendment as an unconstitutionally vague statute.

For many homeless persons, their automobile may be their last major possession—the means by which they can look for work and seek social services. The *1158City of Los Angeles has many options at its disposal to alleviate the plight and suffering of its homeless citizens. Selectively preventing the homeless and the poor from using their vehicles for activities many other citizens also conduct in their cars should not be one of those options.

REVERSED.

Mr. Justice Goldberg

delivered the opinion of the Court.

We are called upon in these two cases to decide the grave and fundamental problem, common to both, of the constitutionality of Acts of Congress which divest an American of his citizenship for “[departing from or remaining outside of the jurisdiction of the United States in time of war or . . . national emergency for the purpose of evading or avoiding training and service” in the Nation’s armed forces.1

*147I. The Facts.

A. Mendoza-Martines — No. 2.

The facts of both cases are not in dispute. Mendoza-Martinez, the appellee in No. 2, was born in this country-in 1922 and therefore acquired American citizenship by birth. By reason of his parentage, he also, under Mexican law, gained Mexican citizenship, thereby possessing dual nationality. In 1942 he departed from this country and went to Mexico solely, as he admits, for the purpose of evading military service in our armed forces. He concedes that he remained there for that sole purpose until November 1946, when he voluntarily returned to this country. In 1947, in the United States District Court for the Southern District of California, he pleaded guilty to and was convicted of evasion of his service obligations in violation of § 11 of the Selective Training and Service Act of 1940.2 He served the imposed sentence of a year and a day. For all that appears in the record, he was, upon his release, allowed to reside undisturbed in this country until *1481953, when, after a lapse of five years, he was served with a warrant of arrest in deportation proceedings. This was premised on the assertion that, by remaining outside the United States to avoid military service after September 27, 1944, when §401 (j) took effect, he had lost his American citizenship. Following hearing, the Attorney General’s special inquiry officer sustained the warrant and ordered that Mendoza-Martinez be deported as an alien. He appealed to the Board of Immigration Appeals of the Department of Justice, which dismissed his appeal.

Thereafter, Mendoza-Martinez brought a declaratory judgment action in the Federal District Court for the Southern District of California, seeking a declaration of his status as a citizen, of the unconstitutionality of § 401 (j), and of the voidness of all orders of deportation directed against him. A single-judge District Court in an unreported decision entered judgment against Mendoza-Martinez in 1955, holding that by virtue of § 401 (j), which the court held to be constitutional, he had lost his nationality by remaining outside the jurisdiction of the United States after September 27, 1944. The Court of Appeals for the Ninth Circuit affirmed the judgment, 238 F. 2d 239. This Court, in 1958, Mendoza-Martinez v. Mackey, 356 U. S. 258, granted certiorari, vacated the judgment, and remanded the cause to the District Court for reconsideration in light of its decision a week earlier in Trop v. Dulles, 356 U. S. 86.

On September 24, 1958, the District Court announced its new decision, also unreported, that in light of Trop §401 (j) is unconstitutional because not based on any “rational nexus . . . between the content of a specific power in Congress and the action of Congress in carrying that power into execution.” On' direct appeal under 28 U. S. C. § 1252, this Court noted probable jurisdiction, 359 U. S. 933, and then of its own motion remanded the cause, this time with permission to the parties to amend *149the pleadings to put in issue the question of whether the facts as determined on the draft-evasion conviction in 1947 collaterally estopped the Attorney General from now claiming that Mendoza-Martinez had lost his American citizenship while in Mexico. Mackey v. Mendoza-Martinez, 362 U. S. 384.

The District Court on remand held that the Government was not collaterally estopped because the 1947 criminal proceedings entailed no determination of Mendoza-Martinez’ citizenship. The court, however, reaffirmed its previous holding that § 401 (j) is unconstitutional, adding as a further basis of invalidity that § 401 (j) is “essentially penal in character and deprives the plaintiff of procedural due process. . . . [T]he requirements of procedural due process are not satisfied by the administrative hearing of the Immigration Service nor in this present proceedings.” 3 The Attorney General’s current appeal is from this decision. Probable jurisdiction was noted on February 20, 1961, 365 U. S. 809. The case was argued last Term, and restored to the calendar for reargument this Term, 369 U. S. 832.

B. Cort — No. 3.

Cort, the appellee in No. 3, is also a native-born American, born in Boston in 1927. Unlike Mendoza-Martinez, he has no dual nationality. His wife and two young children are likewise American citizens by birth. Following receipt of his M. D. degree from the Yale University School of Medicine in 1951, he went to England for the purpose of undertaking a position as a Research Fellow at Cambridge University. He had earlier registered in timely and proper fashion for the draft and shortly before *150his departure supplemented his regular Selective Service registration by registering under the newly enacted Doctors Draft Act.4 In late 1951 he received a series of letters from the American Embassy in London instructing him to deliver his passport to it to be made “valid only for return to the United States.” He did not respond to these demands because, he now says in an affidavit filed in the trial court in this proceeding, “I believed that they were unlawful and I did not wish to subject myself to this and similar forms of political persecution then prevalent in the United States. ... I was engaged in important research and teaching work in physiology and I desired to continue earning a livelihood for my family.” Cort had been a member of the Communist Party while he was a medical student at Yale from 1946 to 1951, except for the academic year 1948-1949 when he was in England. In late 1952, while still in England at Cambridge, he accepted a teaching position for the following academic year at Harvard University Medical School. When, however, the school discovered through further correspondence that he had not yet fulfilled his military obligations; it advised him that it did not regard his teaching position'as essential enough to support his deferment from military service in order to enter upon it. Thereafter, his local draft board in Brookline, Massachusetts, notified him in February 1953 that his request for deferment was denied and that he should report within 30 days for a physical examination either in Brookline or in Frankfurt, Germany. On June 4 and on July 3 the draft board again sent Cort notices to report for a physical examination, the first notice for examination on July 1 in Brookline, and the second for examination within 30 days in Frankfurt. He did not appear at either place, and the board on August 13 ordered him to report for induction on September 14, *1511953. He did not report, and consequently he was indicted in December 1954 for violation of § 12 (a) of the Selective Service Act of 1948 5 by reason of his failure to report for induction. This indictment is still outstanding. His complaint in this action states that he did not report for induction because he believed “that the induction order was not issued in good faith to secure his military services, that his past political associations and present physical disabilities made him ineligible for such service, and that he was being ordered to report back to the United States to be served with a Congressional committee subpoena or indicted under the Smith Act . . . .” Meanwhile, the British Home Office had refused to renew his residence permit, and in mid-1954 he and his family moved to Prague, Czechoslovakia, where he took a position as Senior Scientific Worker at the Cardiovascular Institute. He has lived there since.

In April 1959, his previous United States passport having long since expired, Cort applied at the American Embassy in Prague for a new one. His complaint in this action states that he wanted the passport “in order to return to the United States with his wife and children so that he might fulfill his obligations under the Selective Service laws and his wife might secure medical treatment for multiple sclerosis.” Mrs. Cort received a passport and came to this country temporarily in late 1959, both for purposes of medical treatment and to facilitate arrangements for her husband’s return. Cort’s application, however, was denied on the ground that he had, by his failure to report for induction on September 14, 1953, as ordered, remained outside the country to avoid military service and thereby automatically forfeited his American citizenship by virtue of § 349 (a) (10) of the Immigration *152and Nationality Act of 1952, which had superseded § 401 (j). The State Department’s Passport Board of Review affirmed the finding of expatriation, and the Department’s legal adviser affirmed the decision. Cort, through counsel, thereupon brought this suit in the District Court for the District of Columbia for a declaratory judgment that he is a citizen of the United States, for an injunction against enforcement of §349 (a) (10) because of its unconstitutionality, and for an order directing revocation of the certificate of loss of nationality and issuance of a United States passport to him. Pursuant to Cort’s demand, a three-judge court was convened. The court held that he had remained outside the United States to evade military service, but that § 349 (a) (10) is unconstitutional because “We perceive no substantial difference between the constitutional issue in the Trop case and the one facing us.” It therefore concluded that Cort is a citizen of this country and enjoined the Secretary of State from withholding a passport from Cort on the ground that he is not a citizen and from otherwise interfering with his rights of citizenship. Cort v. Herter, 187 F. Supp. 683.

The Secretary of State appealed directly to this Court, 28 U. S. C. §§ 1252, 1253, which postponed the question of jurisdiction to the hearing of the case on the merits. 365 U. S. 808. The preliminary question of jurisdiction was affirmatively resolved last Term, Rusk v. Cort, 369 U. S. 367, leaving the issue of the validity of § 349 (a) (10) for decision now, after reargument. 369 U. S., at 380.

Before we consider the essential question in these cases, the constitutionality of §§401 (j) and 349 (a)(10), two preliminary issues peculiar to No. 2 must be discussed.

II. The Three-Judge Court Issue.

At the threshold in Mendoza-Martinez’ case is the question whether the proceeding should have been heard by a three-judge District Court convened pursuant to 28 *153U. S. C. § 2282, which requires such a tribunal as a prerequisite to the granting of any “interlocutory or permanent injunction restraining the enforcement, operation or execution of any Act of Congress for repugnance to the Constitution of the United States . . . .” If § 2282 governs this litigation, we are once again faced with the prospect of a remand and a new trial, this time by a three-judge panel. We are, however, satisfied that the case was properly heard by a single district judge, as both parties urge.

In the complaint under which the case was tried the first and second times, Mendoza-Martinez asked for no injunctive relief, and none was granted. In the amended complaint which he filed in 1960 to put in issue the question of collateral estoppel, he added a prayer asking the court to adjudge “that defendants herein are enjoined and restrained henceforth from enforcing” all deportation orders against him. However, it is abundantly clear from the amended trial stipulation which was entered into by the parties and approved by the judge to “govern the course of the trial,” that the issues were framed so as not to contemplate any injunctive relief. The first question was articulated only in terms of whether the Government was “herein estopped by reason of the indictment and conviction of plaintiff for [draft evasion] . . . from denying that the plaintiff is now a national and citizen of the United States.” The second question asked only for a declaration as to whether § 401 (j) was “unconstitutional, either on its face or as applied to the plaintiff herein.” The conclusion that no request for injunctive relief nor even any contemplation of it attended the case as it went to trial is borne out by the total lack of reference to injunctive relief in the District Court’s memorandum opinion, findings of fact and conclusions of law, and judgment. See 192 F. Supp. 1. The relief granted was merely a declaration that the 1944 Amendment “is *154unconstitutional, both on its face and as applied to the plaintiff herein,” and “ [t]hat the plaintiff is now, and ever since the date of his birth has been, a national and citizen of the United States.” Thus, despite the amendment to Mendoza-Martinez’ complaint before the third trial, it is clear that neither the parties nor the judge at any relevant time regarded the action as one in which injunctive relief was material to the disposition of the case. Since no injunction restraining the enforcement of § 401 (j) was at issue, § 2282 was not in terms applicable to require the convening of a three-judge District Court.

Whether an action solely for declaratory relief would under all circumstances be inappropriate for consideration by a three-judge court we need not now decide, for it is clear that in the present case the congressional policy underlying the statute was not frustrated by trial before a single judge. The legislative history of § 2282 and of its complement, § 2281,6 requiring three judges to hear in-junctive suits directed against federal and state legislation, respectively, indicates that these sections were enacted to prevent a single federal judge from being able to paralyze totally the operation of an entire regulatory scheme, either state or federal, by issuance of a broad injunctive order. Section 2281 “was a means of protecting the increasing body of state legislation regulating economic enterprise from invalidation by a conventional suit in equity. . . . The crux of the business is procedural protection against an improvident state-wide doom by a federal court of a state’s legislative policy. This was the aim of Congress . . . .” Phillips v. United States, 312 U. S. 246, *155250-251. Repeatedly emphasized during the congressional debates on § 2282 were the heavy pecuniary costs of the unforeseen and debilitating interruptions in the administration of federal law which could be wrought by a single judge’s order, and the great burdens entailed in coping with harassing actions brought one after another to challenge the operation of an entire statutory scheme, wherever jurisdiction over government officials could be acquired, until a judge was ultimately found who would grant the desired injunction.- 81 Cong. Rec. 479-481, 2142-2143 (1937).

The present action, which in form was for declaratory relief and which in its agreed substance did not contemplate injunctive relief, involves none of the dangers to which Congress was addressing itself. The relief sought and the order entered affected an Act of Congress in a totally noncoercive fashion. There was no interdiction of the operation at large of the statute. It was declared unconstitutional, but without even an injunctive sanction against the application of the statute by the Government to Mendoza-Martinez. Pending review in the Court of Appeals and in this Court, the Government has been free to continue to apply the statute. That being the case, there is here no conflict with the purpose of Congress to provide for the convocation of a three-judge court whenever the operation of a statutory scheme may be immediately disrupted before a final judicial determination of the validity of the trial court’s order can be obtained. Thus there was no reason whatever in this case to invoke the special and extraordinary procedure of a three-judge court. Compare Schneider v. Rusk, post, p. 224, decided this day.

III. The Collateral-Estoppel Issue.

Mendoza-Martinez’ second amended complaint, filed in 1960 pursuant to the suggestion of this Court earlier that year, charged that “the government of the United States *156has admitted the fact of his United States citizenship by virtue of the indictment and judgment of conviction. [in 1947 for draft evasion] . . . and is therefore collaterally estopped now to deny such citizenship . . . The District Court rejected this assertion. Mendoza-Martinez renews it here as an alternative ground for upholding the judgment entered below “That the plaintiff is now, and ever since the date of his birth has been, a national and citizen of the United States.” 192 F. Supp., at 3.

We too reject Mendoza-Martinez’ contention on this point. His argument, stated more fully, is as follows: The Selective Training and Service Act of 1940 applies only to citizens and resident aliens. Both the indictment and the judgment spoke in terms of his having remained in Mexico for the entire period from November 15, 1942, until November 1,1946, when he returned to this country.7 *157For the period from September 27, 1944, when § 401 (j) became effective, until November 1, 1946, he could not have been in violation of our draft laws unless he remained a citizen of the United States, since the draft laws do not apply to nonresident aliens. Therefore, he concludes, the Government must be taken to have admitted that he did not lose his citizenship by remaining outside the country after September 27, 1944, because it charged him with draft evasion for that period as well as for the period preceding that date.

It is true that “as to those matters in issue or points controverted, upon the determination of which the finding or verdict was rendered,” Cromwell v. County of Sac, 94 U. S. 351, 353, the findings in a prior criminal proceeding may estop a party in a subsequent civil action, Emich Motors Corp. v. General Motors Corp., 340 U. S. 558, 568-569, and that the United States may be estopped to deny even an erroneous prior determination of status, United States v. Moser, 266 U. S. 236. However, Mendoza-Martinez’ citizenship status was not at issue in his trial for draft evasion. Putting aside the fact that he pleaded guilty, which in itself may support the conclusion that his citizenship status was not litigated and thereby without more preclude his assertion of estoppel,8 the basic flaw in his argument is in the assertion that he was charged with a continuing violation of the draft laws while he remained in Mexico, particularly after September 27, 1944, the date on which § 401 (j) became effective. He was in fact charged with a violation “on or about November 15, 1942,” because he “did knowingly evade service ... in that he did knowingly depart from *158the United States and go to a foreign country, namely: Mexico, for the purpose of evading service . . . This constituted the alleged violation. The additional language that he “did there remain until on or about November 1, 1946,” was merely surplusage in relation to the substantive offense, although it might, for example, serve a purpose in relation to problems connected with the tolling of the statute of limitations. No language appears charging the elements of violation — knowledge and purpose to evade — in connection with it. The only crime charged is what happened “on or about November 15, 1942,” and conviction thereon, even if it had entailed a finding as to Mendoza-Martinez’ citizenship on that date,9 in nowise estopped the Government with reference to his status after September 27, 1944.

The trial court’s judgment was worded no differently. Mendoza-Martinez was convicted of:

“Having on or about November 15th 1942, knowingly departed from the United States to Mexico, for the purpose of evading service in the land or naval forces of the United States and having remained there until on or about November 1st 1946.”

Again, the language relating to the time during which Mendoza-Martinez remained in Mexico was not tied to the words stating knowledge and purpose to evade service. Thus, the conviction entailed no actual or necessary finding about Mendoza-Martinez’ citizenship status between September 27, 1944, and November 1,1946, and the Government was not estopped from denying his citizenship in the present proceedings.

*159IV. The Constitutional Issues.

A. Basic Principles.

Since the validity of an Act of Congress is involved, we begin our analysis mindful that the function we are now discharging is “the gravest and most delicate duty that this Court is called upon to perform.” Blodgett v. Holden, 275 U. S. 142, 148 (separate opinion of Holmes, J.). This responsibility we here fulfill with all respect for the powers of Congress, but with recognition of the transcendent status of our Constitution.

We deal with the contending constitutional arguments in the context of certain basic and sometimes conflicting principles. Citizenship is a most precious right. It is expressly guaranteed by the Fourteenth Amendment to the Constitution, which speaks in the most positive terms.10 The Constitution is silent about the permissibility of involuntary forfeiture of citizenship rights.11 While it confirms citizenship rights, plainly there are imperative obligations of citizenship, performance of which Congress in the exercise of its powers may constitutionally exact. One of the most important of these is to serve the country in time of war and national emergency. The powers of Congress to require military service for the common defense are broad and far-reachin*160g,’12 for while the Constitution protects against invasions of individual rights, it is not a suicide pact. Similarly, Congress has broad power under the Necessary and Proper Clause to enact legislation for the regulation of foreign affairs. Latitude in this area is necessary to ensure effec-tuation of this indispensable function of government.13

These principles, stemming on the one hand from the preeious nature of the constitutionally guaranteed rights of citizenship, and on the other from the powers of Congress and the related obligations of individual citizens, are urged upon us by the parties here. The Government argues that §§401 (j) and 349 (a)(10) are valid as an exercise of Congress’ power over foreign affairs, of its war power, and of the inherent sovereignty of the Government. Appellees urge the provisions’ invalidity as not within any of the powers asserted, and as imposing a cruel and unusual punishment.

We recognize at the outset that we are confronted here with an issue of the utmost import. Deprivation of citizenship — particularly American citizenship, which is “one of the most valuable rights in the world today,” Report of the President’s Commission on Immigration and Naturalization (1953), 235 — has grave practical consequences. An expatriate who, like Cort, had no other nationality becomes a stateless person — a person who not only has no rights as an American citizen, but no membership in any national entity whatsoever. “Such individuals as do not possess any nationality enjoy, in general, no protection whatever, and if they are aggrieved by a State they have no means of redress, since there is no State which is competent to take up their case. As far as the Law of Na*161tions is concerned, there is, apart from restraints of morality or obligations expressly laid down by treaty . . . no restriction whatever to cause a State to abstain from maltreating to any extent such stateless individuals.” 1 Oppenheim, International Law (8th ed., Lauterpacht, 1955), § 291, at 640.14 The calamity is “[n]ot the loss of specific rights, then, but the loss of a community willing and able to guarantee any rights whatsoever . . . Arendt, The Origins of Totalitarianism (1951), 294. The stateless person may end up shunted from nation to nation, there being no one obligated or willing to receive him,15 or, as in Cort’s case, may receive the dubious sanctuary of a Communist regime lacking the essential liberties precious to American citizenship.16

*162B. The Perez and Trop Cases.

The basic principles here involved, the gravity of the issue, and the arguments bearing upon Congress’ power to forfeit citizenship were considered by the Court in relation to different provisions of the Nationality Act of 1940 in two cases decided on the same day less than five years ago: Perez v. Brownell, 356 U. S. 44, and Trop v. Dulles, 356 U. S. 86.

In Perez, § 401 (e), which imposes loss of nationality for “[vjoting in a political election in a foreign state or participating in an election or plebiscite to determine the sovereignty over foreign territory,” was upheld by a closely divided Court as a constitutional exercise of Congress’ power to regulate foreign affairs. The Court reasoned that since withdrawal of citizenship of Americans who vote in foreign elections is reasonably calculated to effect the avoidance of embarrassment in the conduct of foreign relations, such withdrawal is within the power of Congress, acting under the Necessary and Proper Clause. Since the Court sustained the application of § 401 (e) to denationalize Perez, it did not have to deal with § 401 (j), upon which the Government had also relied, and it expressly declined to rule on the constitutionality of that section, 356 U. S., at 62. There were three opinions written in dissent. The principal one, that of The Chief Justice, recognized “that citizenship may not only be voluntarily renounced through exercise of the right of expatriation but also by other actions in derogation of undivided allegiance to this country,” id., at 68, but concluded that “[t]he mere act of voting in a foreign election, however, without regard to the circumstances attending *163the participation, is not sufficient to show a voluntary abandonment of citizenship,” id., at 78.

In Trop, § 401 (g), forfeiting the citizenship of any American who is guilty of “[djeserting the military or naval forces of the United States in time of war, provided he is convicted thereof by court martial and as the result of such conviction is dismissed or dishonorably discharged . . . ,” was declared unconstitutional. There was no opinion of the Court. The Chief Justice wrote an opinion for four members of the Court, concluding that § 401 (g) was invalid for the same reason that he had urged as to § 401 (e) in his dissent in Perez, and that it was also invalid as a cruel and unusual punishment imposed in violation of the Eighth Amendment. Justice Brennan conceded that it is “paradoxical to justify as constitutional the expatriation of the citizen who has committed no crime by voting in a Mexican political election, yet find unconstitutional a statute which provides for the expatriation of a soldier guilty of the very serious crime of desertion in time of war,” 356 U. S., at 105. Notwithstanding, he concurred because “the requisite rational relation between this statute and the war power does not appear . . . ,” id., at 114. Justice Frankfurter, joined by three other Justices, dissented on the ground that § 401 (g) did not impose punishment at all, let alone cruel and unusual punishment, and was within the war powers of Congress.

C. Sections Jfil (;) and S/fi (a) (10) as Punishment.

The present cases present for decision the constitutionality of a section not passed upon in either Perez or Trop — § 401 (j), added in 1944, and its successor and present counterpart, §349 (a) (10) of the Immigration and Nationality Act of 1952. We have come to the conclusion that there is a basic question in the present cases, *164the answer to which obviates a choice here between the powers of Congress and the constitutional guarantee of citizenship. That issue is whether the statutes here, which automatically — -without prior court or administrative proceedings — impose forfeiture of citizenship, are essentially penal in character, and consequently have deprived the appellees of their citizenship without due process of law and without according them the rights guaranteed by the Fifth and Sixth Amendments, including notice, confrontation, compulsory process for obtaining witnesses, trial by jury, and assistance of counsel. This issue was not relevant in Trop because, in contrast to §§ 401 (j) and 349 (a) (10), § 401 (g) required conviction by court-martial for desertion before forfeiture of citizenship could be inflicted. In Perez the contention that § 401 (e) was penal in character was impliedly rejected by the Court’s holding, based on legislative history totally different from that underlying §§ 401 (j) and 349 (a) (10), that voting in a political election in a foreign state “is regulable by Congress under its power to deal with foreign affairs.” 356 U. S., at 59. Compare Dent v. West Virginia, 129 U. S. 114; Hawker v. New York, 170 U. S. 189; Flemming v. Nestor, 363 U. S. 603. Indeed, in Trop The Chief Justice observed that “Section 401 (j) decrees loss of citizenship without providing any semblance of procedural due process whereby the guilt of the draft evader may be determined before the sanction is imposed . . . ,” 356 U. S., at 94, and Justice Frankfurter in dissent alluded to the due process overtones of the requirement in § 401 (g) of prior conviction for desertion by court-martial, id., at 116-117.

It is fundamental that the great powers of Congress to conduct war and to regulate the Nation’s foreign relations are subject to the constitutional requirements of due *165process.17 The imperative necessity for safeguarding these rights to procedural due process under the gravest of emergencies has existed throughout our constitutional history, for it is then, under the pressing exigencies of crisis, that there is the greatest temptation to dispense with fundamental constitutional guarantees which, it is feared, will inhibit governmental action. “The Constitution of the United States is a law for rulers and people, equally in war and in peace, and covers with the shield of its protection all classes of men, at all times, and under all circumstances.” Ex parte Milligan, 4 Wall. 2, 120-121.18 The rights guaranteed by the Fifth and Sixth Amendments are “preserved to every one accused of crime who is not attached to the army, or navy, or militia in actual service.” Id., at 123.19 “[I]f society is disturbed by civil commotion — if the passions of men are aroused and the restraints of law weakened, if not disregarded — these safeguards need, and should receive, the watchful care of those intrusted with the guardianship of the Constitution and laws. In no other way can we transmit to posterity unimpaired the blessings of liberty, consecrated by the sacrifices of the Revolution.” Id., at 124.

We hold §§ 401 (j) and 349 (a)(10) invalid because in them Congress has plainly employed the sanction of deprivation of nationality as a punishment — for the offense of leaving or remaining outside the country to evade mili*166tary service — without affording the procedural safeguards guaranteed by the Fifth and Sixth Amendments.20 Our forefathers “intended to safeguard the people of this country from punishment without trial by duly constituted courts. . . . And even the courts to which this important function was entrusted were commanded to stay their hands until and unless certain tested safeguards were observed. An accused in court must be tried by an impartial jury, has a right to be represented by counsel, [and] must be clearly informed of the charge against him . . . .” United States v. Lovett, 328 U. S. 303, 317. See also Chambers v. Florida, 309 U. S. 227, 235-238.

As the Government concedes, §§ 401 (j) and 349 (a) (10) automatically strip an American of his citizenship, with concomitant deprivation “of all that makes life worth living,” Ng Fung Ho v. White, 259 U. S. 276, 284-285, whenever a citizen departs from or remains outside the jurisdiction of this country for the purpose of evading his military obligations. Conviction for draft evasion, as *167Cort’s case illustrates, is not prerequisite to the operation of this sanction.21 Independently of prosecution, forfeiture of citizenship attaches when the statutory set of facts develops. It is argued that the availability after the fact of administrative and judicial proceedings, including the machinery the Court approved last Term in Rusk v. Cort, 369 U. S. 367, to contest the validity of the sanction meets the measure of due process. But the legislative history and judicial expression with respect to every congressional enactment relating to the provisions in question dating back to 1865 establish that forfeiture of citizenship is a penalty for the act of leaving or staying outside the country to avoid the draft. This being so, the Fifth and Sixth Amendments mandate that this punishment cannot be imposed without a prior criminal trial and all its incidents, including indictment, notice, confrontation, jury trial, assistance of counsel, and compulsory process for obtaining witnesses. If the sanction these sections impose is punishment, and it plainly is, the procedural safeguards required as incidents of a criminal prosecution are lacking. We need- go no further.

*168The punitive nature of the sanction here is evident under the tests traditionally applied to determine whether an Act of Congress is penal or regulatory in character, even though in other cases this problem has been extremely difficult and elusive of solution. Whether the sanction involves an affirmative disability or restraint,22 whether it has historically been regarded as a punishment,23 whether it comes into play only on a finding of scienter, 24 whether its operation will promote the traditional aims of punishment — retribution and deterrence,25 whether the behavior to which it applies is already a crime,26 whether an alternative purpose to which it may *169rationally be connected is assignable for it,27 and whether it appears excessive in relation to the alternative purpose, assigned28 are all relevant to the inquiry, and may often point in differing directions. Absent conclusive evidence of congressional intent as to the penal nature of a statute, these factors-must be considered in relation to the statute on its face. Here, although we are convinced that application of these criteria to the face of the statutes supports the conclusion that they are punitive, a detailed examination along such lines is unnecessary, because the objective manifestations of congressional purpose indicate conclusively that the provisions in question can only be interpreted as punitive.29 A study of the history of the predecessor of §401 (j), which “is worth a volume of logic,” New York Trust Co. v. Eisner, 256 U. S. 345, 349, coupled with a reading of Congress’ reasons for enacting § 401 (j), compels a conclusion that the statute’s primary function is to serve as an additional penalty for *170a special category of draft evader.30 Compare Trop v. Dulles, supra, 356 U. S., at 107-110 (Brennan, J., concurring).

1. The Predecessor Statute and Judicial Construction.

The subsections here in question have their origin in part of a Civil War “Act to amend the several Acts heretofore passed to provide for the Enrolling and Calling out the National Forces, and for other Purposes.” Act of March 3,1865,13 Stat. 487. Section 21 of that Act, dealing with deserters and draft evaders, was in terms punitive, providing that “in addition to the other lawful penalties of the crime of desertion,” persons guilty thereof “shall be deemed and taken to have voluntarily relinquished and forfeited their rights of citizenship and their rights to become citizens . . . and all persons who, being duly enrolled, shall depart the jurisdiction of the district in which he is enrolled, or go beyond the limits of the United States, with intent to avoid any draft into the *171military or naval service, duly ordered, shall be liable to the penalties of this section.” 31

The debates in Congress in 1865 confirm that the use of punitive language in § 21 was not accidental. The section as originally proposed inflicted loss of rights of citizenship only on deserters. Senator Morrill of Maine proposed amending the section to cover persons who leave the country to avoid the draft, stating, “I do not see why the same principle should not extend to those who leave the country to avoid the draft.” Cong. Globe, 38th Cong., 2d Sess. 642 (1865). This “same principle” was punitive, because Senator Morrill was also worried that insofar as the section as originally proposed “provides for a penalty” to be imposed on persons who had theretofore deserted, there was question “whether it is not an ex post facto law, whether it is not fixing a penalty for an act already done.” Ibid. Senator Johnson of Maryland attempted to allay Senator Morrill’s concern by explaining that

“the penalties are not imposed upon those who have deserted, if nothing else occurs, but only on those who have deserted and who shall not return within sixty days. The crime for which the punishment is inflicted is made up of the fact of an antecedent desertion, and a failure to return within sixty days. It is clearly within the power of Congress.” Ibid.

This explanation satisfied the Senate sufficiently so that they accepted the section, with Senator Morrill’s amendment, although Senator Hendricks of Indiana-made one last speech in an effort to convince his colleagues of the bill’s ex post facto nature and, even apart from that, of the excessiveness of the punishment, particularly as applied to draft evaders:

“It seems to me to be very clear that this section proposes to punish desertions which have already *172taken place, with a penalty which the law does not already prescribe. In other words it is an ex post facto criminal law which I think we cannot pass. . . . One of the penalties known very well to the criminal laws of the country is the denial of the right of suffrage and the right to hold offices of trust or profit.

“It seems to me this objection to the section is very clear, but I desire to suggest further that this section punishes desertions that may hereafter take place in the same manner, and it is known to Senators that one desertion recently created is not reporting when notified of the draft. ... I submit to Senators that it is a horrible thing to deprive a man of his citizenship, of that which is his pride and honor, from the mere fact that he has been unable to report upon the day specified after being notified that he has been drafted. Certainly the punishment for desertion is severe enough. It extends now from the denial of pay up to death; that entire compass is given for the punishment of this offense. Why add this other? It cannot do any good.” Id., at 643.

In the House, the motion of New York’s Representative Townsend to strike the section as a “despotic measure” which would “have the effect to deprive fifty thousand, and I do not know but one hundred thousand, people of their rights and privileges,” was met by the argument of Representative Schenck of Ohio, the Chairman of the Military Committee, that “Here is a penalty that is lawful, wise, proper, and that should be added to the other lawful penalties that now exist against deserters.” Id., at 1155. After Representative Wilson of Iowa proposed an amendment, later accepted and placed in the enacted version of the bill, extending the draft-evasion portion to apply to persons leaving “the district in which they are enrolled” in addition to those leaving the country, Representative J. C. Allen of Illinois raised the ex post facto *173objection to the section as a whole. Id., at 1155-1156. Representative Schenck answered him much as Senator Johnson had replied in the Senate:

“The gentleman from Illinois [Mr. J. C. Allen] misapprehends this section from not having looked carefully, as I think, into its language. He thinks it retroactive. It is not so. It does not provide for punishing those who have deserted in their character of deserters acquired by having gone before the passage of the law, but of those only, who, being deserters, shall not return and report themselves for duty within sixty days. If the gentleman looks at the language of the section, he will find that we have carefully avoided making it retroactive. We give those who have deserted their country and their flag sixty days for repentance and return.

“Mr. J. C. ALLEN. Will not the infliction of this penalty on those who have failed to return to the Army be an additional penalty that did not exist at the time they deserted?

“Mr. SCHENCK. Yes, sir.

“Mr. J. C. ALLEN. Does not that make the law retroactive?

“Mr. SCHENCK. They are deserters now. We take them up in their present status and character as deserters, and punish them for continuing in that character. The gentleman refers to lawyers here. I believe he is a good lawyer himself. Does he not know that if a man steals a horse and runs away with it to the next county it is a continual act of larceny until he delivers up the horse?” Id., at 1156.

The significance of these debates is, as these excerpts plainly show, that while there was a difference in both Houses as to whether the statute would be an ex post facto law, there was agreement among all the speakers on both *174sides of that issue, as well as on both sides of the merits of the bill generally, that deprivation of rights of citizenship for leaving the country to evade the draft was a “penalty” and “punishment” for a “crime” and an “offense” and a violation of a “criminal law.”

A number of state court judicial decisions rendered shortly after the Civil War lend impressive support to the conclusion that the predecessor of §§ 401 (j) and 349 (a) (10), §21 of the 1865 statute, was a criminal statute imposing an additional punishment for desertion and draft evasion. The first and most important of these was Huber v. Reily, 53 Pa. 112 (1866), in which, as in most of the cases which followed,32 the plaintiff had brought an action against the election judge of his home township, alleging that the defendant had refused to receive his ballot on the ground that plaintiff was a deserter and thereby disenfranchised under § 21, and that such refusal was wrongful because § 21 was unconstitutional. The asserted grounds of invalidity were that § 21 was an ex post jacto law, that it was an attempt by Congress to regulate suffrage in the States and therefore outside Congress’ sphere of power, and that it proposed to inflict pains and penalties without a trial and conviction, and was therefore prohibited by the Bill of Rights. In an opinion by Justice Strong, later a member of this Court, the Pennsylvania Supreme Court first characterized the statute in a way which compelled discussion of the asserted grounds of unconstitutionality:

“The Act of Congress is highly penal. It imposes forfeiture of citizenship and deprivation of the rights of citizenship as penalties for the commission of a crime. Its avowed purpose is to add to the penalties which the law had previously affixed to the offence *175of desertion from the military or naval service of the United States, and it denominates the additional sanctions provided as penalties.” 53 Pa., at 114-115.

It then answered the ex post facto argument as it had been answered on the floor of Congress, that the offense could as well be in the continued refusal to render service as in the original desertion. The second contention was met with the statement that “The enactment operates upon an individual offender, punishes him for violation of the Federal law by deprivation of his citizenship of the United States, but it leaves each state to determine for itself whether such an individual may be a voter. It does no more than increase the penalties of the law upon the commission of crime.” Id., at 116. “The third objection,” the court continued, “would be a very grave one if the act does in reality impose pains and penalties before and without a conviction by due process of law.” Id., at 116-117. The court then summarized the protections guaranteed by the Fifth and Sixth Amendments, and concluded that it was not consistent with these rights to empower a “judge of elections or a board of election officers constituted under state laws ... to adjudge the guilt or innocence of an alleged violator of the laws of the United States.” Id., at 117. However, the court decided that since the penalty contemplated by § 21 “is added to what the law had previously enacted to be the penalty of desertion, as imprisonment is sometimes added to punishment by fine,” it must have been intended “that it should be incurred in the same way, and imposed by the same tribunal that was authorized to impose the other penalties for the offence.” Id., at 119. “[T]he forfeiture which it prescribes, like all other penalties for desertion, must be adjudged to the convicted person, after trial by a court-martial, and sentence approved. For the conviction and sentence of such a court there can be no substitute.” Id., *176at 120. (Emphasis in original.) Accordingly, since the plaintiff had not been so convicted, the court held that he was not disenfranchised.

Subsequent state court decisions in the post-Civil War period followed Huber v. Reily, both in result and reasoning. State v. Symonds, 57 Me. 148 (1869); Severance v. Healey, 50 N. H. 448 (1870); Gotcheus v. Matheson, 58 Barb. (N. Y.) 152 (1870); McCafferty v. Guyer, 59 Pa. 109 (1868).

Ultimately and significantly, in Kurtz v. Moffitt, 115 U. S. 487, a case dealing with the question whether a city police officer had the power to arrest a military deserter, this Court recognized both the nature of the sanction imposed by § 21 and the attendant necessity of procedural safeguards, approvingly citing the above decisions:

“The provisions of §§ 1996 and 1998, which re-enact the act of March 3, 1865, ch. 79, § 21, 13 Stat. 490, and subject every person deserting the military service of the United States to additional penalties, namely, forfeiture of all rights of citizenship, and disqualification to hold any office of trust or profit, can only take effect upon conviction by a court martial, as was clearly shown by Mr. Justice Strong, when a judge of the Supreme Court of Pennsylvania, in Huber v. Reily, 53 Penn. St. 112, and has been uniformly held by the civil courts as well as by the military authorities. State v. Symonds, 57 Maine, 148; Severance v. Healey, 50 N. H. 448; Goetcheus v. Matthewson, 61 N. Y. 420; Winthrop’s Digest of Judge Advocate General’s Opinions, 225.” 115 U. S., at 501-502.

Section 21 remained on the books unchanged, except for being distributed in the Revised Statutes as §§ 1996 and 1998, until 1912, when Congress re-enacted it with an amendment making it inapplicable to peacetime violations *177and giving the President power to mitigate or remit punishment previously imposed on peacetime violators, Act of August 22, 1912, 37 Stat. 356. The legislative history of that amendment is also instructive for our present inquiry. The discussion in both Houses had reference only to the penalties as operative on deserters, no doubt because there was no peacetime draft to evade, but since the 1865 statute dealt without distinction with both desertion and leaving the jurisdiction to evade, there is no reason to suppose the discussion quoted below to be any less applicable to the latter type of misconduct. The House Committee Report, H. R. Rep. No. 335, 62d Cong., 2d Sess. (1912), which was quoted in its entirety in the Senate Committee Report, S. Rep. No. 910, 62d Cong., 2d Sess. 3-6 (1912), stated that “In addition to the service penalty imposed by the court-martial, the law, as it now stands, imposes the further and most drastic punishment of loss of rights of citizenship .... There are in the United States to-day thousands of men who are literally men without a country and their numbers will be constantly added to until the drastic civil-war measure which adds this heavy penalty to an already severe punishment imposed by military law, is repealed.” H. R. Rep. No. 335, supra, at 2. In reporting the bill out of the Committee on Naval Affairs, Representative Roberts of Massachusetts, its author, stated that “the bill now under consideration is intended to remove one of the harshest penalties that can be imposed upon a man for an offense, to wit, the loss of rights of citizenship. . . . [S]uch a drastic penalty was entirely too severe to be imposed upon an American citizen in time of peace.” He detailed the penalties meted out by court-martial for desertion, and then referred to the “additional penalty of loss of citizenship,” which, he concluded, is “a barbarous punishment.” 48 Cong. Rec. 2903 (1912). Senator Bristow of Kansas, a member of his chamber’s Committee on Military Affairs, *178also referred in discussing the bill to the forfeiture of rights of citizenship as a “penalty,” and said that there is no reason why a peacetime offender should be “punished so severely.” 48 Cong. Rec. 9542 (1912).

A somewhat similar amendment had been passed by both Houses of Congress in 1908 but vetoed by the President.33 The House Committee Report on that occasion, H. R. Rep. No. 1340, 60th Cong., 1st Sess. (1908), consisted mainly of a letter from the Secretary of the Navy to the Congress, and of his annual report. In both documents he referred to loss of citizenship as a “punishment,” and as one of the “penalties” for desertion. Representative Roberts spoke in 1908, as he was to do once more in 1912, of the “enormity of the punishment” and the “horrible punishment,” and said, “Conviction itself under *179the existing law forfeits citizenship. That is the monstrosity of the law.” 43 Cong. Rec. Ill (1908). The entire discussion, id., at 110-114, was based on the premise that loss of citizenship is a punishment for desertion, the point at issue, as in 1912, being whether it was too severe a punishment for peacetime imposition. At one point Representative Roberts said, “Loss of citizenship is a punishment,” to which Representative Hull of Iowa replied, “Certainly.” Id., at 114.

Section 504 of the Nationality Act of 1940, 54 Stat. 1172, repealed the portion of the 1865 statute which dealt with flight from the jurisdiction to avoid the draft. However, in connection with the provision governing loss of citizenship for desertion, which was enacted as § 401 (g) and declared unconstitutional in Trop v. Dulles, supra, the President’s committee of advisers reported that the provisions of the 1865 Act had been “distinctly penal in character,” and concluded that “They must, therefore, be construed strictly, and the penalties take effect only upon conviction by a court martial.” 34 Codification of the Nationality Laws of the United States, 76th Cong., 1st Sess. 68 (Comm. Print 1939). Section 401 (g) was therefore worded so that loss of nationality could only occur upon conviction for desertion by court-martial. When, however, § 401 (j) was enacted in 1944, no such procedural safeguards were built in. See Trop v. Dulles, supra, at 93-94. Thus, whereas for Justice Brennan concurring in Trop the conclusion that expatriation under § 401 (g) was punishment was “but the beginning of critical inquiry,” 356 U. S., at 110, a similar conclusion with reference to § § 401 (j) and 349 (a) (10) is sufficient to sustain the holding that they are unconstitutional.

*1802. The Present Statutes.

The immediate legislative history of § 401 (j) confirms the conclusion, based upon study of the earlier legislative and judicial history,35 that it is punitive in nature. The language of the section was, to begin with, quite obviously patterned on that of its predecessor, an understandable fact since the draft of the bill was submitted to the Congress by Attorney General Biddle along with a letter to Chairman Russell of the Senate Immigration Committee, in which the Attorney General referred for precedent to the 1912 reenactment of the 1865 statute. This letter, which was the impetus for the enactment of the bill, was quoted in full text in support of it in both the House and Senate Committee Reports, H. R. Rep. No. 1229, 78th Cong., 2d Sess. 2-3 (1944); S. Rep. No. 1075, 78th Cong., 2d Sess. 2 (1944), and is set out in the margin.36 The *181Senate Report stated that it “fully explains the purpose of the bill.” S. Rep. No. 1075, supra, at 1. The letter was couched entirely in terms of an argument that citizens who had left the country in order to escape military serv*182ice should be dealt with, and that loss of citizenship was a proper way to deal with them. There was no reference to the societal good that would be wrought by the legislation, nor to any improvement in soldier morale or in the conduct of war generally that would be gained by the passage of the statute. The House Committee Report and the sponsors of the bill endorsed it on the same basis. The report referred for support to the fact that the FBI files showed “over 800 draft delinquents” in the El Paso area alone who had crossed to Mexico to evade the draft. H. R. Rep. No. 1229, supra, at 2. The obvious inference to be drawn from the report, the example it contained, and the lack of mention of any broader purpose is that Congress was concerned solely with inflicting effective retribution upon this class of draft evaders and, no doubt, on others similarly situated. Thus, on the floor of the House, Representative Dickstein of New York, the Chairman of the House Committee on Immigration and Naturalization, explained the bill solely as a means of dealing with “draft dodgers who left this country knowing that there was a possibility that they might be drafted in this war and that they might have to serve in the armed forces . . . .” He implied that the bill was necessary to frustrate their “idea of evading military service and of returning after the war is over, and taking their old places *183in our society.” 90 Cong. Rec. 3261 (1944). Senator Russell, who was manager of the bill as well as Chairman of the Senate Immigration Committee, explained it in similar terms:

“Certainly those who, having enjoyed the advantages of living in the United States, were unwilling to serve their country or subject themselves to the Selective Service Act,- should be penalized in some measure. . . . Any American citizen who is convicted of violating the Selective Service Act loses his citizenship. This bill would merely impose a similar penalty on those who are not subject to the jurisdiction of our courts, the penalty being the same as would result in the case of those who are subject to the jurisdiction of our courts.” 90 Cong. Rec. 7629 (1944).37

The Senate and House debates, together with Attorney General Biddle’s letter, brought to light no alternative purpose to differentiate the new statute from its predecessor. Indeed, as indicated, the Attorney General’s letter specifically relied on the predecessor statute as precedent for this enactment, and both the letter and the debates, consistent with the character of the predecessor statute, referred to reasons for the enactment of the bill which were fundamentally retributive in nature. When all of these considerations are weighed, as they must be, in the context of the incontestibly punitive nature of the predecessor statute, the conclusion that § 401 (j) was itself dominantly punitive becomes inescapable. The legislative history of § 349 (a) (10) of the Immigration and Nationality Act of 1952, which re-enacted § 401 (j), adds *184nothing to disturb that result.38 Our conclusion from the legislative and judicial history is, therefore, that Congress in these sections decreed an additional punishment for the crime of draft avoidance in the special category of cases wherein the evader leaves the country. It cannot do this without providing the safeguards which must attend a criminal prosecution.39

V. Conclusion.

It is argued that our holding today will have the unfortunate result of immunizing the draft evader who has left the United States from having to suffer any sanction against his conduct, since he must return to this country before he can be apprehended and tried for his crime. The compelling answer to this is that the Bill of Rights which we guard so jealously and the procedures it guarantees are not to be abrogated merely because a guilty man may escape prosecution or for any other expedient reason. Moreover, the truth is that even without being expatriated, the evader living abroad is not in a position to assert the vast majority of his component rights as an American citizen. If he wishes to assert those rights in any real sense he must return to this country, and by doing that he will subject himself to prosecution. In fact, *185while he is outside the country evading prosecution, the United States may, by proper refusal to exercise its largely discretionary power to afford him diplomatic protection,40 decline to invoke its sovereign power on his behalf. Since the substantial benefits of American citizenship only come into play upon return to face prosecution, the draft evader who wishes to exercise his citizenship rights will inevitably come home and pay his debt, which within constitutional limits Congress has the power to define. This is what Mendoza-Martinez did, what Cort says he is willing to do, and what others have done.41 Thus our holding today does not frustrate the effective handling of the problem of draft evaders who leave the United States.42

*186We conclude, for the reasons stated, that §§ 401 (j) and 349 (a) (10) are punitive and as such cannot constitutionally stand, lacking as they do the procedural safeguards which the Constitution commands.43 We recognize that draft evasion, particularly in time of war, is a heinous offense, and should and can be properly punished. Dating back to Magna Carta, however, it has been an abiding principle governing the lives of civilized men that “no freeman shall be taken or imprisoned or disseised or outlawed or exiled . . . without the judgment of his peers or by the law of the land . . . .” 44 What we hold is only that, in keeping with this cherished tradition, punishment cannot be imposed “without due process of law.” Any lesser holding would ignore the constitutional mandate upon which our essential liberties depend. Therefore the judgments of the District Courts in these cases are

Affirmed.

Mr. Justice Douglas and Mr. Justice Black, while joining the opinion of the Court, adhere to the views éx-pressed in the dissent of Mr. Justice Douglas, in which Mr. Justice Black joined, in Perez v. Brownell, 356 U. S. 44, 79, that Congress has no power to deprive a person of the citizenship granted the native-born by § 1, cl. 1, of the Fourteenth Amendment.

*187Mr. Justice Brennan,

concurring.

I join the Court’s opinion because I fully agree with the Court’s conclusion that Congress has here attempted to employ expatriation as a penal sanction in respect of behavior deemed inimical to an objective whose pursuit is within its assigned powers, and with the reasoning by which that conclusion is reached. So too, I agree that Congress is constitutionally debarred from so employing the drastic, the truly terrifying remedy of expatriation, certainly where no attempt has been made to apply the full panoply of protective safeguards which the Constitution requires as a condition of imposing penal sanctions. However, I deem it appropriate to elaborate somewhat the considerations which impel me to agree with the Court.

This Court has never granted, the existence in Congress of the power to expatriate except where its exercise was intrinsically and peculiarly appropriate to the solution of serious problems -inevitably implicating nationality. We have recognized the entanglements which may stem from dual allegiance, and have twice sustained statutes which provided for loss of American citizenship upon the deliberate assumption of a foreign attachment. Mackenzie v. Hare, 239 U. S. 299; Savorgnan v. United States, 338 U. S. 491. We have recognized that participation by American nationals in the internal politics of foreign states could dangerously prejudice our diplomacy, and have allowed the use of expatriation as a uniquely potent corrective which precludes recriminations by disowning, at the moment of his provocative act, him who might otherwise be taken as our spokesman or our operative. Perez v. Brownell, 356 U. S. 44. The instant cases do not require me to resolve some felt doubts of the correctness of Perez, which I joined. For the Court has never held that expatriation was to be found in Congress’ *188arsenal of common sanctions, available for no higher purpose than to curb undesirable conduct, to exact retribution for it, and to stigmatize it.

I.

In Trop v. Dulles, 356 U. S. 86, we had before us § 401 (g) of the Nationality Act of 1940, which imposed loss of American nationality following conviction of deserting the armed forces in time of war. We held that statute unconstitutional. Three of my Brethren joined in the opinion of The Chief Justice, who analyzed the case in terms equally applicable to the cases at bar. That plurality opinion in Trop noted that the congressional power to which expatriation under § 401 (g) was said to be relevant was the “war power.” It concluded that expatriation under § 401 (g) could have no value in furtherance of the war power except as a sanction, to deter or punish desertion; that expatriation so employed was “punishment” within the meaning of the Eighth Amendment; and that such punishment was unconstitutional because cruel and unusual.1

My concurring views in Trop, separately expressed, were akin to those of the plurality. I shared the view that expatriation could have been employed in § 401 (g) only as a sanction, and I considered this an insufficient predicate for its use — which I believed allowable only where some affirmative and unique relationship to policy was apparent. My premise was the simple and fundamental one that legislation so profoundly destructive of individual rights must keep within the limits *189of palpable reason and rest upon some modicum of discoverable necessity. I was unable to conclude that § 401 (g) met that elementary test. It was evident that recognizable achievement of legitimate congressional purposes through the expatriation device was at best remote; and that far more promising alternative methods existed and had, in fact, been employed.

My Brother Stewart attempts to distinguish Trop along two fronts: He argues that expatriation is not here employed as “punishment” in the constitutional sense so that the reasoning of the Trop plurality has no application; and he argues that, the question of punishment aside, expatriation as here employed is a uniquely necessary device not falling within the rationale of my views separately expressed in Trop.

My Brother Stewart discerns in § 401 (j) 2 an affirmative instrument of policy and not simply a sanction which must be classed as “punishment.” The policy objective is thought to be the maintenance of troop morale; a threat to that objective is thought to be the spectacle of persons escaping a military-service obligation by flight; and expatriation of such persons is sustained as a demonstrative counter to that threat. To my mind that would be “punishment” in the purest sense; it would be naked vengeance. Such an exaction of retribution would not lose that quality because it was undertaken to maintain morale. Indeed, it is only the significance of expatriation as retribution which could render it effective to boost morale — -the purpose which, to the dissent, removes expatriation as here used from the realm of the punitive. I do not perceive how expatriation so employed would differ analytically from the stocks or the rack. Because *190such devices may be calculated to shore up the convictions of the law-abiding by demonstrating that the wicked will not go unscathed, they would not, by the dissent’s view, be punitive or, presumably, reachable by the Eighth Amendment.3 I cannot agree to any such proposition, and I see no escape from the conclusion that § 401 (j), before us today, is identical in purpose to § 401 (g) and is quite as “punitive” as was that statute, which we condemned in Trop.

The dissent finds other distinctions between this case and Trop, quite apart from its untenable position that § 401 (j) is not punitive. It is said that flight from the country to escape the draft, in contrast with desertion, could never be a mere technical offense equivocal in its implications for the loyalty of the offender. But the unshakable fear of physical stress or harm, the intellectual or moral aversion to combat, and the mental aberration which may result in flight are no more inconsistent with underlying loyalty than was Trop’s unauthorized abandonment of his post.4 Again, it is suggested that the *191element of cumulation of punishments which helped expose the futility of expatriation in Trop is missing here, because § 401 (j), unlike 1401(g), becomes operative without a prior conviction, and applies only in the case of flight beyond our borders. But the Mendoza-Martinez case, in its collateral estoppel issue, prominently displays what would in any case be obvious — that expatriation under § 401 (j) is cumulative with criminal sanctions for draft evasion, for those sanctions apply to fugitives equally as much as to sedentary violators.5

Nor can Trop rationally be distinguished on the ground that the application of § 401 (j) only to fugitives proves that it was designed to fill a void necessarily left by the ordinary criminal draft-evasion sanctions. The point, as I understand it, is that the ordinary sanctions cannot be brought to bear against a fugitive who declines to come home; but he can be expatriated while he remains abroad, without having to be brought before a tribunal and formally proceeded against. The special virtue of expatriation, it appears, is that it maybe accomplished in absentia.

*192Aside from the denial of procedural due process, which the Court rightly finds in the scheme, the surface appeal of the argument vanishes upon closer scrutiny.

It simply is not true that expatriation provides an instrumentality specially necessary for imposing the congressional will upon fugitive draft evaders. Our statutes now provide severe criminal sanctions for the behavior in question. The fugitive can return only at the cost of suffering these punishments; the only way to avoid them is to remain away. As to any draft delinquent for whom the prospect of this dilemma would not itself pose a recognizable, formidable deterrent, I fail to see how the addition of expatriation could enhance the effect at all.6 Nor can expatriation affect the fugitive who will not return to be punished — for whom it is thought to be specially designed. For that individual has, ex hypothesi, determined on his own to stay away and so cannot be affected by the withdrawal of his right to return. The sting of the measure is felt only by those like Mendoza-Martinez, who have already returned and been punished, and those like Cort, who desire to return and be punished — those, in other words, as to whom expatriation is patently cumulative with other sanctions. As to the unregenerate fugitive whom it is particularly thought to reach, expatriation is but a display of congressional displeasure. I cannot agree that it is within the power of Congress so to express its displeasure with those who will *193not return as to destroy the rights and the status of those who have demonstrated their underlying attachment to this country by coming home.

It is apparent, then, that today’s cases are governed by Trop no matter which of the two controlling opinions is consulted. Expatriation is here employed as “punishment,” cruel and unusual here if it was there. Nor has expatriation as employed in these cases any more rational or necessary a connection with the war power than it had in Trop.

II.

Me. Justice Stewaet’s dissent would sustain § 401 (j) as a permissible exercise of the “war power.” The appellants in these cases, on the other hand, place their main reliance on the “foreign affairs power.” The dissent summarizes the appellants’ arguments under this heading but does not purport to pass on them. Because of my conviction that § 401 (j) is unconstitutional no matter what congressional power is invoked in its support, I find it necessary to deal with the foreign affairs arguments advanced by the appellants.

Initially, I note that the legislative history as expounded by the dissent fails to reveal that Congress was mindful of any foreign affairs problem to be corrected by the statute. The primary purpose seems to me to have been retributive, the secondary purpose deterrent; and even the morale-boosting purpose discerned by the dissent has nothing to do with foreign affairs. While the obvious fact that Congress was not consciously pursuing any foreign affairs objective may not necessarily preclude reliance on that power as a ground of constitutionality, it does render such reliance initially questionable.

Proceeding to the appellants’ arguments, one encounters first the suggestion that a fugitive draft evader “can easily cause international complications” while he remains *194an American citizen, because the United States cannot exercise control over him while he is on foreign soil.

Such a “problem,” obviously, exists equally with respect to any fugitive from American justice, and cannot be thought confined to draft evaders. Yet it is only fugitive draft evaders who are expatriated. It is, therefore, impossible to agree that Congress was acting on any such inherently unlikely premise as that expatriation was necessary so as to avoid responsibility for those described by § 401 (j).

But, contend the appellants, §401 (j) is designed to prevent embroilments as well as embarrassments. During wartime, it is argued, our Government would very likely feel impelled to demand of foreign havens the return of our fugitive draft evaders; and such a demand might seriously offend a “host” country, leading to embroilment. The transparent weakness of this argument — its manifest inconsistency — must be immediately apparent. Surely the United States need not disable itself from making injudicious demands in order to restrain itself from doing so. The argument rests on the possibility that there may be an urgent need to secure a fugitive’s return. If that is so, a demand must be made with its attendant risk of embroilment. If expatriating the fugitive makes a demand impossible, it also forever defeats the objective— his return — which would have impelled the demand in the first place. If recapturing fugitives may ever be urgently necessary, it is obvious that automatic expatriation could only be directly opposed to our interest — which requires that the Government be free to choose whether or not to make the demand, in light of all the attendant circumstances.

The appellants have still another argument. It is that whereas the Government is under an obligation to seek the return of the fugitive as long as he remains a citizen, by terminating citizenship “Congress has eliminated at *195the outset any further claim that this country would have to the services of these individuals, and has removed all basis for further demands upon them . . . This simply is not so. Expatriation may have no effect on a continuing military-service obligation.7 And it is incontrovertible that the power to punish the initial draft-evasion offense continues although citizenship has meanwhile become forfeit. The Government has so argued in addressing itself to the collateral estoppel issue in Mendoza-Martinez. I cannot understand how any obligation to apprehend can be other than coextensive with the power to punish. The Government cannot have it both ways in the same case.

III.

The appellants urge that, wholly apart from any explicit congressional power, § 401 (j) may be sustained as an exercise of a power inherent in United States sovereignty. My Brethren who would uphold the statute have not' adverted to this possibility except, as I shall point out, as they have adopted in passing certain related arguments.

Preliminarily, it is difficult to see what is resolved by the assertion that sovereignty implies a power to expatriate. That proposition may be admitted and yet have no bearing on the problem facing the Court.

For, under our Constitution, only a delimited portion of sovereignty has been assigned to the Government of *196which Congress is the legislative arm. To say that there inheres in United States sovereignty the power to sever the tie of citizenship does not answer the inquiry into whether that power has been granted to Congress. Any argument that it has been so delegated which eschews reference to the constitutional text must, it appears, make its appeal to some sense of the inevitable fitness of things. The contentions here fall far short of any such standard.

It is too simple to suggest that it is fitting that Congress be empowered either to extinguish the citizenship of one who refuses to perform the “ultimate duty” of rising to the Government’s defense in time of crisis. I pause to note that for this Court to lend any credence whatever to such a criterion — as the dissent would, see pp. 214-215, infra> — is fraught with the most far-reaching consequences. For if Congress now should declare that a refusal to pay taxes, to do jury duty, to testify, to vote, is no less an abnegation of ultimate duty — or an implied renunciation of allegiance — than a refusal to perform military service, I am unable to perceive how this Court, on the dissent’s view, could presume to gainsay such a judgment. But the argument is not saved even by a willingness to accept these consequences. There really is no way to distinguish between the several failures of a citizen’s duty I have just enumerated, or to explain why evasion of military service should be visited with this specially harsh consequence, except to recognize that the latter defection is palpably more provocative than the others. But, as I have argued in another context, when conduct is singled out of a class for specially adverse treatment simply because it is specially provocative, there is no escaping the conclusion that punishment is being administered. See Flemming v. Nestor, 363 U. S. 603, 635-640 (dissenting opinion). Pursuit of the “ultimate duty” concept, then, simply reaffirms my conviction that this case is indistinguishable from Trop.

*197The appellants, however, argue that it is fitting that Congress be empowered to extinguish the citizenship of one who not only refuses to perform his duty, but who also “repudiates his wider obligation as a citizen to submit to this country’s jurisdiction and authority” by fleeing the country in order to escape that duty. It is, once again, difficult to see how this flight-repudiation theory can be confined to draft evasion. Every fugitive from United States justice repudiates American authority over him in equal measure. If the difference lies in the quality of the act of draft evasion, then we are back once again to punishment.

The appellants assert that “[a] government which cannot exert force to compel a citizen to perform his lawful [Government’s emphasis] duty is, to that extent, not sovereign as to him.” The apparent corollary is that congressionally imposed expatriation is, under such circumstances, in effect declaratory of a change in status which has already occurred. But the Government is far from conceding its lack of authority over a fugitive draft evader. It informs us that “the federal government has the power to order our citizens abroad to return, for any lawful purpose,” citing Blackmer v. United States, 284 U. S. 421. And, in any event, the argument proves far too much, for it would justify expatriation of any American abroad for any reason who would, equally with persons covered by § 401 (j), be outside our Government’s power to compel the performance of duty.

Me. Justice Harlan,

whom Mr. Justice Clark joins, dissenting.

I agree with and join in Parts I, II, III, and IV of my Brother Stewart’s opinion, leading to the conclusion that §401 (j) of the Nationality Act of 1940, applicable in No. 2 {Mendoza), is constitutional. I also agree with his conclusion that, for the same reasons, the substantive *198provisions of § 349 (a) (10) of the 1952 Act, applicable in No. 3 {Cort), are constitutional. I disagree, however, with his view that the evidentiary presumption contained in § 349 (a) (10) is unconstitutional. I am content to state my reasons in summary form.

1. As I read the opinion below in the Cort case I do not think the District Court relied on the § 349 (a) (10) presumption.1 This view is fortified by several considerations: (i) the constitutionality of the presumption was attacked in Cort’s complaint and was briefed by both sides in the District Court; (ii) the text of the presumption itself was set forth in the opinion of the District Court (187 F. Supp., at 684) at only a page or two before the extract quoted in the margin (note 1); and (iii) in these *199circumstances it is difficult to believe that the lower court, composed of three experienced judges, either inadvertently ignored the presumption or upheld its validity sub silentio. The more likely conclusion is that finding the evidence sufficient without the aid of the presumption, the lower court saw no need for reaching a second constitutional issue.

So viewing the District Court’s opinion, I think the evidence was quite sufficient under the “clear, unequivocal, and convincing” standard of Schneiderman v. United States, 320 U. S. 118, 135, to support the finding below that Cort had remained abroad for the purpose of evading military service.2

*2002. In addition, I see nothing constitutionally wrong with this presumption either on its face or as related to this case. Similar presumptions have been consistently sustained in criminal statutes, where the standard of proof is certainly no less stringent than in denationalization cases. See, e. g., Yee Hem v. United States, 268 U. S. 178; Casey v. United States, 276 U. &. 413; Hawes v. Georgia, 258 U. S. 1; cf. Fong Yue Ting v. United States, 149 U. S. 698. As regards the requirement that .there must be a “rational connection between the fact proved and the ultimate fact presumed,” Tot v. United States, 319 U. S. 463, 467, this presumption is surely a far cry *201from that held constitutionally invalid in the Tot case.3 And since we are concerned here only with the presumption as applied in this instance (if indeed it was in fact applied below or must now be resorted to in this Court), it is no answer to suggest that in other instances application of the presumption might be unconstitutional.

Thus whether or not the § 349 (a) (10) presumption is involved in the Cort case, I believe that the order of denationalization there, as well as in the Mendoza case, should be upheld.4

Mr. Justice Stewart,

with whom Mr. Justice White joins, dissenting.

The Court’s opinion is lengthy, but its thesis is simple: (1) The withdrawal of citizenship which these statutes provide is “punishment.” (2) Punishment cannot constitutionally be imposed except after a criminal trial and conviction. (3) The statutes are therefore unconstitu*202tional. As with all syllogisms, the conclusion is inescapable if the premises are correct. But I cannot agree with the Court's major premise — that the divestiture of citizenship which these statutes prescribe is punishment in the constitutional sense of that term.1

I.

Despite the broad sweep of some of the language of its opinion, the Court as I understand it does not hold that involuntary deprivation of citizenship is inherently and always a penal sanction — requiring the safeguards of a criminal trial. Such a determination would overrule at least three decisive precedents in this Court.

Nearly 50 years ago the Court held that Congress had constitutional power to denationalize a native-born citizen who married a foreigner but continued to reside here. Mackenzie v. Hare, 239 U. S. 299. The Court there explicitly rejected the argument “that the citizenship of plaintiff was an incident to her birth in the United States, and, under the Constitution and laws of the United States, it became a right, privilege and immunity which could not be taken away from her except as a punishment for crime or by her voluntary expatriation.” 239 U. S., at 308. The power of Congress to/ denationalize a native-born citizen, without a criminal trial, was reaffirmed in Savorgnan v. United States, 338 U. S. 491. And less than five years ago, in Perez v. Brownell, 356 U. S. 44, the Court again upheld this congressional power in an opinion which unambiguously rejected the notion, advanced in *203that case by the dissenters,2 that the Mackenzie and Savorgnan decisions stand only for the proposition that citizenship may be voluntarily relinquished or abandoned either expressly or by conduct. In short, it has been established for almost 50 years that Congress under some circumstances may, without providing for a criminal trial, make expatriation the consequence of the voluntary conduct of a United States citizen, irrespective of the citizen’s subjective intention to renounce his nationality, and irrespective too of his awareness that denationalization will be the result of his conduct.3

II.

The position taken by the Court today is simply that, unlike the statutes involved in Mackenzie, Savorgnan and Perez, the statutes at issue in the present case employ deprivation of citizenship as a penal sanction. In support of this position, the Court devotes many pages of its opinion to a discussion of a quite different law, enacted in 1865, amended in 1912, and repealed in 1940. That law4 provided for forfeiture of the “rights of citizenship” as an additional penalty for deserters from the armed forces and for enrolled draftees who departed from their district or from the United States “to avoid any draft into the military or naval service, duly ordered . . . .” That statute, as the Court correctly says, “was in terms *204punitive,” and I agree with the Court that the statute’s legislative history, as well as subsequent judicial decisions construing it, makes it clear that the law was punitive- — • imposing additional punishment upon those convicted of either of the offenses mentioned.5

In these cases, however, we have before us statutes which were enacted in 1944 and 1952, respectively. In construing these statutes, I think nothing is to be gained from the legislative history of a quite different law enacted by a quite different Congress in 1865, nor from the reports of still another Congress which amended that law in 1912. Unlike the 1865 law, the legislation at issue in the cases before us is not “in terms punitive.” And there is nothing in the history of this legislation which persuades me that these statutes, though not in terms penal, nonetheless embody a purpose of the Congresses which enacted them to impose criminal punishment without the safeguards of a criminal trial.

Unlike the two sections of the Nationality Act of 1940 which were in issue in Perez v. Brownell6 and Trop v. Dulles,7 § 401 (j) did not have its genesis in the Cabinet Committee’s draft code which President Roosevelt submitted to Congress in 1938.8 Indeed, § 401 (j) was the product of a totally different environment — the experience of a nation engaged in a global war.

On February 16, 1944, Attorney General Biddle addressed a letter to the Chairman of the Senate Immigra*205tion Committee, calling attention to circumstances which had arisen after the institution of the draft in World War II, and suggesting the legislation which subsequently became § 401 (j). The Attorney General’s letter stated in part:

“I invite your attention to the desirability of enacting legislation which would provide (1) for the expatriation of citizens of the United States who in time of war or during a national emergency leave the United States or remain outside thereof for the purpose of evading service in the armed forces of the United States and (2) for the exclusion from the United States of aliens who leave this country for the above-mentioned purpose.

“Under existing law a national of the United States, whether by birth or by naturalization, becomes expatriated by operation of law if he (1) obtains naturalization in .a foreign state; (2) takes an oath of allegiance to a foreign country; (3) serves in the armed forces of a foreign state if he thereby acquires the nationality of such foreign state; (4) accepts employment under a foreign state for which only nationals of such state are eligible; (5) votes in a political election in a foreign state or participates in an election or plebiscite to determine the sovereignty over foreign territory; (6) makes a formal renunciation of nationality before a diplomatic or consular officer of the United States in a foreign state; (7) deserts from the armed forces of the United States in time of war and is convicted thereof by a court martial; or (8) is convicted of treason (U. S. C., title 8, sec. 801). Machinery is provided whereby a person who is denied any right or privilege of citizenship on the ground that he has become expatriated may secure a judicial determination of his status; and if he is outside of the United States he is entitled to a *206certificate of identity which permits him to enter and remain in the United States until his status has been determined by the courts (Nationality Act of 1940, sec. 503; U. S. C., title 8, sec. 903).

“The files of this Department disclose that at the present time there are many citizens of the United States who have left this country for the purpose of escaping service in the armed forces. While such persons are liable to prosecution for violation of the Selective Service and Training Act of 1940, if and when they return to this country, it would seem proper that in addition they should lose their United States citizenship. Persons who are unwilling to perform their duty to their country and abandon it during its time of need are much less worthy of citizenship than are persons who become expatriated on any of the existing grounds.

“Accordingly, I recommend the enactment of legislation which would provide (1) for the expatriation of citizens of the United States who in time of war or during a national emergency leave the United States or remain outside thereof for the purpose of evading service in the armed forces of the United States and (2) for the exclusion from the United States of aliens who leave this country for that purpose. Any person who may be deemed to have become expatriated by operation of the foregoing provision would be entitled to have his status determined by the courts pursuant to the above-mentioned section of the Nationality Act of 1940.” 9

The bill was passed unanimously by both the House and the Senate, and became Public Law No. 431 of the Seventy-eighth Congress. Neither the committee reports nor the limited debate on the measure in Congress *207adds any substantial gloss to the legislative action.10 And the legislative history of § 349 (a) (10) of the Immigration and Nationality Act of 1952, the statute directly involved in the second of the two cases now before us, *208gives no additional illumination as to the purpose of the Eighty-second Congress, since the substantive provisions of that statute were but a recodification of § 401 (j) of the 1940 Act.11

The question of whether or not a statute is punitive ultimately depends upon whether the disability it imposes is for the purpose of vengeance or deterrence, or whether the disability is but an incident to some broader regulatory objective. See Cummings v. Missouri, 4 Wall. 277, 320, 322; United States v. Lovett, 328 U. S. 303, 308-312; *209 Trop v. Dulles, 356 U. S., at 107-109. See generally, Flemming v. Nestor, 363 U. S. 603, 613-617; cf. De Veau v. Braisted, 363 U. S. 144,160; Communist Party v. Subversive Activities Control Board, 367 U. S. 1, 83-88. In commenting on the nature of this kind of inquiry, the Court said in Flemming v. Nestor, “We observe initially that only the clearest proof could suffice to establish the unconstitutionality of a statute on such a ground. Judicial inquiries into Congressional motives are at best a hazardous matter, and when that inquiry seeks to go behind objective manifestations it becomes a dubious affair indeed. Moreover, the presumption of constitutionality with which this enactment, like any other, comes to us forbids us lightly to choose that reading of the statute’s setting which will invalidate it over that which will save it.” 363 U. S., at 617.

In the light of the standard enunciated in Nestor, I can find no clear proof that the prime purpose of this legislation was punitive. To be sure, there is evidence that the deterrent effect of the legislation was considered. Moreover, the attitude of some members of Congress toward those whom the legislation was intended to reach was obviously far from neutral. But the fact that the word “penalty” was used by an individual Senator in the congressional debates is hardly controlling. As The Chief Justice has so wisely remarked, “How simple would be the tasks of constitutional adjudication and of law generally if specific problems could be solved by inspection of the labels pasted on them!” 12

It seems clear to me that these putative indicia of punitive intent are far overbalanced by the fact that this legislation dealt with a basic problem of wartime morale reaching far beyond concern for any individual affected. The legislation applies only to those who have left this *210country or remained outside of it for the purpose of avoiding the draft. Congress can reasonably be understood to have been saying that those who flee the country for such express purposes do more than simply disobey the law and avoid the imposition of criminal sanctions. They disassociate themselves entirely from their nation, seeking refuge from their wartime obligations under the aegis of another sovereign. Congress could reasonably have concluded that the existence of such a group, who voluntarily and demonstrably put aside their United States citizenship “for the duration,” could have an extremely adverse effect upon the morale and thus the war effort not only of the armed forces, but of the millions enlisted in the defense of their nation on the civilian front. During the consideration of § 401 (j) in Congress there were repeated references to the expectation that fugitive draft evaders then living abroad would return to this country after the war to resume citizenship and to enjoy the fruits of victory. The effect upon wartime morale of the known existence of such a group, while perhaps not precisely measurable in terms of impaired military efficiency, could obviously have been considered substantial. Denationalization of this class of voluntary expatriates was a rational way of dealing with this problem by removing its visible cause. In light of this broader purpose, I cannot find, as the Court does, that § 401 (j) was motivated primarily by the desire to wreak vengeance upon those individuals who fled the country to avoid military service. Rather, the statute seems to me precisely the same kind of regulatory measure, rational and efficacious, which this Court upheld against similar objections in Perez v. Brownell, supra. 13

*211III.

For the reasons stated, I cannot find in the terms of these statutes or in their legislative history anything close to the “clearest proof” that the basic congressional purpose was to impose punishment. But that alone does not answer the constitutional inquiry in these cases. As with any other exercise of congressional power, a law which imposes deprivation of citizenship, to be constitutionally valid, must bear a rational relationship to an affirmative power possessed by Congress under the Constitution. The appellants submit that in enacting this legislation, Congress could rationally have been drawing on any one of three sources of recognized constitutional power: the implied power to enact legislation for the effective conduct of foreign affairs; the express power to wage war, to raise armies, and to provide for the common defense; and the inherent attributes of sovereignty.

The appellants argue that this legislation, like the statutory provision sustained in Perez v. Brownell, supra, has a direct relationship to foreign affairs. They point out that international complications could arise if this country attempted to effect the return of citizen draft evaders by requests to a foreign sovereign which that nation might be unwilling to grant. The appellants insist that the possibility of international embroilments resulting from problems caused by fugitive draft evaders is not fanciful, pointing to the background of international incidents preceding the War of 1812, and the long history, later in the nineteenth century, of this country's involvement with other nations over the asserted liability of our naturalized citizens to military obligations imposed by their native countries.14 Expatriation of those who leave or remain *212away from the United States with draft evasion as their purpose, the appellants say, might reasonably be attributed to a congressional belief that this was the only practical way to nip these potential international problems in the bud. Compare. Perez v. Brownell, 356 U. S., at 60; Trop v. Dulles, 356 U. S., at 106 (concurring opinion).

In the view I take of this case, it is unnecessary to pursue further an inquiry as to whether the power to regulate foreign affairs could justify denationalization for the conduct in question. For I think it apparent that Congress in enacting the statute was drawing upon another power, broad and far reaching.

A basic purpose of the Constitution was to “provide for the common defence.” To that end, the Framers expressly conferred upon Congress a compendium of powers which have come to be called the “war power.” 15 Responsive to the scope and magnitude of ultimate national need, the war power is “the power to wage war successfully.” See Charles Evans Hughes, War Powers under the Constitution, 42 A. B. A. Rep. 232, 238.

It seems to me evident that Congress was drawing upon this power when it enacted the legislation before us. To be sure, the underlying purpose of this legislation can *213hardly be refined to the point of isolating one single, precise objective. The desire to end a potential drain upon this country’s military manpower was clearly present in the minds of the legislators and would itself have constituted a purpose having sufficient rational nexus to the exercise of the war power. Indeed, there is no more fundamental aspect of this broad power than the building and maintaining of armed forces sufficient for the common defense. Selective Draft Law Cases, 245 U. S. 366; see Falbo v. United States, 320 U. S. 549. But, in any event, the war power clearly supports the objective of removing a corrosive influence upon the morale of a nation at war. As the Court said in Hirabayashi v. United States, 320 U. S. 81, 93, the war power “extends to every matter, and activity so related to war as substantially to affect its conduct and progress. The power is not restricted to the winning of victories in the field and the repulse of enemy forces. It embraces every phase of the national defense, including the protection of war materials and the members of the armed forces from injury and from the dangers which attend the rise, prosecution and progress of war.” See Lichter v. United States, 334 U. S. 742.

This legislation is thus quite different from the statute held invalid in Trop v. Dulles, supra. In that case there were not five members of the Court who were able to find the “requisite rational relation” between the war power of Congress and § 401 (g) of the 1940 Act imposing denationalization upon wartime deserters from the armed forces. As the concurring opinion pointed out, the statute was “not limited in its effects to those who desert in a foreign country or who flee to another land.” 356 U. S., at 107. Indeed, “The Solicitor General acknowledged that forfeiture of citizenship would have occurred if the entire incident had transpired in this country.” 356 U. S., at 92. It was emphasized that conduct far short of disloyalty could technically constitute the military offense *214of desertion, 356 U. S., at 112, 113, and that the harshness of denationalization for conduct so potentially equivocal was “an important consideration where the asserted power to expatriate has only a slight or tenuous relation to the granted power.” 356 U. S., at 110.

The legislation now before us, on the other hand, is by its terms completely inapplicable to those guilty of draft evasion who have remained in the United States; it is exclusively aimed at those, whether or not ever criminally convicted, who have gone to or remained in another land to escape the duty of military service. Moreover, the conduct which the legislation reaches could never be equivocal in nature, but is always and clearly a “refusal to perform this ultimate duty of American citizenship.” Trop v. Dulles, 356 U. S., at 112 (concurring opinion).

IV.

There is one more point to be made as to the substantive provisions of the legislation before us in these cases. Previous decisions have suggested that congressional exen cise of the power to expatriate may be subject to a further constitutional restriction — a limitation upon the kind of activity which may be'made the basis of denationalization. Withdrawal of citizenship is a drastic measure. Moreover, the power to expatriate endows government with authority to define and to limit the society which it represents and to which it is responsible.

This Court has never held that Congress’ power to expatriate may be used unsparingly in every area in which it has general power to act. Our previous decisions upholding involuntary denationalization all involved conduct inconsistent with undiluted allegiance to this country. But I think the legislation at issue in these cases comes so clearly within the compass of those decisions as to make unnecessary in this case an inquiry as to *215what the ultimate limitation upon the expatriation power may be.

The conduct to which this legislation applies, involving not only the attribute of flight or absence from this country in time of war or national emergency, but flight or absence for the express purpose of evading the duty of helping to defend this country, amounts to an unequivocal and conspicuous manifestation of nonallegiance, whether considered objectively or subjectively. Ours is a tradition of the citizen soldier. As this Court has said, “[T]he very conception of a just government and its duty to the citizen includes the reciprocal obligation of the citizen to render military service in case of need and the right to compel it.” Selective Draft Law Cases, 245 U. S. 366, at 378. It is hardly an improvident exercise of constitutional power for Congress to disown those who have disowned this Nation in time of ultimate need.

V.

For the reasons stated, I believe the substantive provisions of § 401 (j) of the 1940 Act and of § 349 (a) (10) of the 1952 Act are constitutionally valid. In addition to its substantive provisions, however, § 349 (a) (10) declares:

“For the purposes of this paragraph failure to comply with any provision of any compulsory service laws of the United States shall raise the presumption that the departure from or absence from the United States was for the purpose of evading or avoiding training and service in the military, air, or naval forces of the United States.”

I think the evidentiary presumption which the statute creates is clearly invalid, and that it fatally infected the administrative determination that Joseph Henry Cort had lost his citizenship.

*216The District Court did not mention this statutory presumption, and it is, therefore, impossible to know how much the court relied upon it, if at all. Indeed, the District Court’s attention in this case was oriented primarily towards the issue of its jurisdiction and the basic issue of the constitutionality of the substantive provisions of § 349 (a) (10). In view of its holding that § 349 (a) (10) is unconstitutional, the court understandably did not give exhaustive attention to the factual issues presented, devoting but a single short paragraph to the question of whether Cort’s conduct had brought him within the statute. 187 F. Supp., at 686.

But it is clear that the final reviewing agency in the State Department relied heavily upon this presumption in determining that Cort had lost his citizenship. The Board of Review on the Loss of Nationality, in its memorandum affirming the initial administrative determination that Cort had lost his citizenship, stated that “[b]y failing to comply with the notices sent to him by his local board, Dr. Cort brought upon himself the presumption mentioned in Section 349 (a) (10), that his continued absence from the United States was for the purpose of evading or avoiding training and service in the military, air, or naval forces of the United States. Even if the Board should consider that the presumption could be overcome by showing that a person remained abroad for a purpose other than to avoid the military service, the evidence in Dr. Cort’s case, taken as a whole, does not show that he remained abroad for a purpose other than to avoid being drafted.” (Emphasis added.) One of the Board’s specific findings was “that Dr. Cort has not overcome the presumption raised in the last sentence of Section 349 (a) (10) of the Immigration and Nationality Act.”

As was said in Speiser v. Randall, 357 U. S. 513, at 520-521, “it is commonplace that the outcome of a law*217suit — and hence the vindication of legal rights — depends more often on how the factfinder appraises the facts than on a disputed construction of a statute or interpretation of a line of precedents. Thus the procedures by which the facts of the case are determined assume an importance fully as great as the validity of the substantive rule of law to be applied. And the more important the rights at stake the more important must be the procedural safeguards surrounding those rights.”

The presumption created by § 349 (a) (10) is wholly at odds with the decisions of the Court which hold that in cases such as this a heavy burden is upon the Government to prove an act of expatriation by clear, convincing, and unequivocal evidence. Gonzales v. Landon, 350 U. S. 920; Nishikawa v. Dulles, 356 U. S. 129. This standard commands that “evidentiary ambiguities are not to be resolved against the citizen.” Nishikawa v. Dulles, 356 U. S., at 136.

Without pausing to consider whether this evidentiary standard is a constitutional one, it is clear to me that the statutory presumption here in question is constitutionally invalid because there is insufficient “rational connection between the fact proved and the ultimate fact presumed.” Tot v. United States, 319 U. S. 463, 467. “A statute creating a presumption that is arbitrary or that operates to deny a fair opportunity to repel it violates the due process clause of the Fourteenth Amendment.” Manley v. Georgia, 279 U. S. 1, 6. A federal, statute which creates such a presumption is no less violative of Fifth Amendment due process. “Mere legislative fiat may not take the place of fact in the determination of issues involving life, liberty or property.” Ibid. It is “essential that there shall be some rational connection between the fact proved and the ultimate fact presumed, and that the inference of one fact from proof of another shall not be *218so unreasonable as to be a purely arbitrary mandate.” Mobile, J. & K. C. R. Co. v. Turnipseed, 219 U. S. 35, 43. Cf. Speiser v. Randall, supra.

The failure of a person abroad to comply with notices sent by his draft board would obviously be relevant evidence in determining whether that person had gone or remained abroad for the purpose of avoiding military service. But the statute goes much further. It creates a presumption of an expatriating act from failure to comply with “any provision of any compulsory service laws” by a citizen abroad, regardless of the nature of the violations and regardless of the innocence of his purpose in originally leaving the United States. The various compulsory service laws of the United States contain a multitude of provisions, many of them technical or relatively insignificant. To draw from the violation of a single such provision a presumption of expatriation, with its solemn consequences, is, I think, to engage in irrationality so gross as to be constitutionally impermissible.16

It is clear from the record in this case that Cort’s sole purpose in leaving the United States in 1951 was to accept a position as a Research Fellow at the University of Cambridge, England. The record also makes clear that in 1946 Cort was called up under the Selective Service law, physically examined, and classified as 4F because of physical disability. The record further shows that Cort voluntarily registered under the Doctors Draft Act, making special arrangements with his draft board to do so in advance of the effective date for registration under the statute, a few days before he left for Europe. Cort filed an affidavit in which he swore that it was his belief, *219in the light of his physical disability, that the induction order which he received in England was not issued in good faith to secure his military service, but that its purpose instead was to force him to return to the United States to be investigated by the House Committee on Un-American Activities or prosecuted under the Smith Act. He has made repeated efforts to arrange with Selective Service officials for the fulfillment, albeit belatedly, of his military obligations, if any, and in 1959 his wife came to the United States and met with officials of the Selective Service system for that purpose. The very reason he applied in Prague for a United States passport was, as he swore, so that he could return to the United States in order to respond to the indictment for draft evasion now pending against him in Massachusetts and to fulfill his Selective Service obligations, if any. When Cort applied in Prague for a passport, the American Consul there, who interviewed him, stated his opinion in writing that he had no reason to disbelieve Cort’s sworn statement that he had not remained outside the United States to avoid military service.17 I mention this evidence as disclosed by the present record only to indicate why I think a new administrative hearing freed from the weight of the statutory presumption is in order, not to imply any prejudgment of what I think the ultimate administrative decision should be.

In No. 3, Rusk v. Cort, I would vacate the judgment of the District Court and remand the case with instructions to declare null and void the certificate of loss of nationality *220issued to Cort by the Secretary of State, so that upon Cort’s renewed application for a passport, an administrative hearing could be had, free of the evidentiary presumption of § 349 (a) (10). In the event that such administrative proceedings should result in a finding that Cort had lost his United States citizenship, he would be entitled to a de novo judicial hearing18 in which the Government would have the burden of proving an act of expatriation by clear, convincing, and unequivocal evidence. Gonzales v. Landon, 350 U. S. 920; Nishikawa v. Dulles, 356 U. S. 129.

In No. 2, Kennedy v. Mendoza-Martinez, I would reverse the judgment of the District Court.

*350Justice Thomas

delivered the opinion of the Court.

In 1994, Kansas enacted the Sexually Violent Predator Act, which establishes procedures for-the civil commitment of persons who, due to a “mental abnormality” or a “personality disorder,” are likely to engage in “predatory acts of sexual violence.” Kan. Stat. Ann. § 59-29a01 et seq. (1994). The State invoked the Act for the first time to commit Leroy Hendricks, an inmate who had a long history of sexually molesting children, and who was scheduled for release from prison shortly after the Act became law. - Hendricks challenged his commitment on, inter alia, “substantive” due process, double jeopardy, and ex post facto grounds. The Kansas Supreme Court invalidated the Act, holding that its precommitment condition of a “mental abnormality” did not satisfy what the court perceived to be the “substantive” due process requirement that involuntary civil commitment must be predicated on a finding of “mental illness.” In re Hendricks, 259 Kan. 246, 261, 912 P. 2d 129, 138 (1996). The State of Kansas petitioned for certiorari. Hendricks subsequently filed a cross-petition in which he reasserted his federal double jeopardy and ex post facto claims. We granted certiorari on both the petition and the cross-petition, 518 U. S. 1004 (1996), and now reverse the judgment below.

I

A

The Kansas Legislature enacted the Sexually Violent Predator Act (Act) in 1994 to grapple with the problem of managing repeat sexual offenders.1 Although Kansas al*351ready had a statute addressing the involuntary commitment of those defined as “mentally ill,” the legislature determined that existing civil commitment procedures were inadequate to confront the risks presented by “sexually violent predators.” In the Act’s preamble, the legislature explained:

“[A] small but extremely dangerous group of sexually violent predators exist who do not have a mental disease or defect that renders them appropriate for involuntary treatment pursuant to the [general involuntary civil commitment statute]_ In contrast to persons appropriate for civil commitment under the [general involuntary civil commitment statute], sexually violent predators generally have anti-social personality features which are unamenable to existing mental illness treatment modalities and those features render them likely to engage in sexually violent behavior. The legislature further finds that sexually violent predators’ likelihood of engaging in repeat acts of predatory sexual violence is high. The existing involuntary commitment procedure ... is inadequate to address the risk these sexually violent predators pose to society. The legislature further finds that the prognosis for rehabilitating sexually violent predators in a prison setting is poor, the treatment needs of this population are very long term and the treatment modalities for this population are very different than the traditional treatment modalities for people appropriate for commitment under the [general involuntary civil commitment statute].” Kan. Stat. Ann. § 59-29a01 (1994).

As a result, the legislature found it necessary to establish “a civil commitment procedure for the long-term care and *352treatment of the sexually violent predator.” Ibid. The Act defined a “sexually violent predator” as:

“any person who has been convicted of or charged with a sexually violent offense and who suffers from a mental abnormality or personality disorder which makes the person likely to engage in the predatory acts of sexual violence.” § 59-29a02(a).

A “mental abnormality” was defined, in turn, as a “congenital or acquired condition affecting the emotional or volitional capacity which predisposes the person to commit sexually violent offenses in a degree constituting such person a menace to the health and safety of others.” § 59-29a02(b).

As originally structured, the Act’s civil commitment procedures pertained to: (1) a presently confined person who, like Hendricks, “has been convicted of a sexually violent offense” and is scheduled for release; (2) a person who has been “charged with a sexually violent offense” but has been found incompetent to stand trial; (3) a person who has been found “not guilty by reason of insanity of a sexually violent offense”; and (4) a person found “not guilty” of a sexually violent offense because of a mental disease or defect. § 59— 29a03(a), §22-3221 (1995).

The initial version of the Act, as applied to a currently confined person such as Hendricks, was designed to initiate a specific series of procedures. The custodial agency was required to notify the local prosecutor 60 days before the anticipated release of a person who might have met the Act’s criteria. §59-29a03. The prosecutor was then obligated, within 45 days, to decide whether to file a petition in state court seeking the person’s involuntary commitment. § 59— 29a04. If such a petition were filed, the court was to determine whether “probable cause” existed to support a finding that the person was a “sexually violent predator” and thus eligible for civil commitment. Upon such a determination, transfer of the individual to a secure facility for professional evaluation would occur. § 59-29a05. After that evaluation, *353a trial would be held to determine beyond a reasonable doubt whether the individual was a sexually violent predator. If that determination were made, the person would then be transferred to the custody of the Secretary of Social and Rehabilitation Services (Secretary) for “control, care and treatment until such time as the person’s mental abnormality or personality disorder has so changed that the person is safe to be at large.” § 59-29a07(a).

In addition to placing the burden of proof upon the State, the Act afforded the individual a number of other procedural safeguards. In the case of an indigent person, the State was required to provide, at public expense, the assistance of counsel and an examination by mental health care professionals. § 59-29a06. The individual also received the right to present and cross-examine' witnesses, and the opportunity to review documentary evidence presented by the State. § 59-29a07.

Once an individual was confined, the Act required that “[t]he involuntary detention or commitment . . . shall conform to constitutional requirements for care and treatment.” § 59-29a09. Confined persons were afforded three different avenues of review: First, the committing court was obligated to conduct an annual review to determine whether continued detention was warranted. §59-29a08. Second, the Secretary was permitted, at any time, to decide that the confined individual’s condition had so changed that release was appropriate, and could then authorize the person to petition for release. § 59-29al0. Finally, even without the Secretary’s permission, the confined person could at any time file a release petition. §59-29all. If the court found that the State could no longer satisfy its burden under the initial commitment standard, the individual would be freed from confinement.

B

In 1984, Hendricks was convicted of taking “indecent liberties” with two 13-year-old boys. After serving nearly 10 years of his sentence, he was slated for release to a halfway *354house. Shortly before his scheduled release, however, the State filed a petition in state court seeking Hendricks’ civil confinement as a sexually violent predator. On August 19, 1994, Hendricks appeared before the court with counsel and -moved to dismiss the petition on the grounds that the Act violated various federal constitutional provisions. Although the court reserved ruling on the Act’s constitutionality, it concluded that there was probable cause to support a finding that Hendricks was a sexually violent predator, and therefore ordered that he be evaluated at the Larned State Security Hospital.

Hendricks subsequently requested a jury trial to determine whether he qualified as a sexually violent predator. During that trial, Hendricks’ own testimony revealed a chilling history of repeated child sexual molestation and abuse, beginning in 1955 when he exposed his genitals to two young girls. At that time, he pleaded guilty to indecent exposure. Then, in 1957, he was convicted of lewdness involving a young girl and received a brief jail sentence. In 1960, he molested two young boys while he worked for a carnival. After serving two years in prison for that offense, he was paroled, only to be rearrested for molesting a 7-year-old girl. Attempts were made to treat him for his sexual deviance, and in 1965 he was considered “safe to be at large,” and was discharged from a state psychiatric hospital. App. 139-144.

Shortly thereafter, however, Hendricks sexually assaulted another young boy and girl — he performed oral sex on the 8-year-old girl and fondled the 11-year-old boy. He was again imprisoned in 1967, but refused to participate in a sex offender treatment program, and thus remained incarcerated until his parole in 1972. Diagnosed as a pedophile, Hendricks entered into, but then abandoned, a treatment program. He testified that despite having received professional help for his pedophilia, he continued to harbor sexual desires for children. Indeed, soon after his 1972 parole, Hendricks began to abuse his own stepdaughter and stepson. He forced the children to engage in sexual activity with him *355over a period of approximately four years. Then, as noted above, Hendricks was convicted of “taking indecent liberties” with two adolescent boys after he attempted to fondle them. As a result of that conviction, he was once again imprisoned, and was serving that sentence when he reached his conditional release date in September 1994.

Hendricks admitted that he had repeatedly abused children whenever he was not confined. He explained that when he “get[s] stressed out,” he “can’t control the urge” to molest children. Id., at 172. Although Hendricks recognized that his behavior harms children, and he hoped he would not sexually molest children again, he stated that the only sure way he could keep from sexually abusing children in the future was “to die.” Id., at 190. Hendricks readily agreed with the state physician’s diagnosis that he suffers from pedophilia and that he is not cured of the condition; indeed, he told the physician that “treatment is bull-.” Id., at 153, 190.2

The jury unanimously found beyond a reasonable doubt that Hendricks was a sexually violent predator. The trial court subsequently determined, as a matter of state law, that pedophilia qualifies as a “mental abnormality” as defined by *356the Act, and thus ordered Hendricks committed to the Secretary’s custody.

Hendricks appealed, claiming, among other things, that application of the Act to him violated the Federal Constitution’s Due Process, Double Jeopardy, and Ex Post Facto Clauses. The Kansas Supreme Court accepted Hendricks’ due process claim. 259 Kan., at 261, 912 P. 2d, at 138. The court declared that in order to commit a person involuntarily in a civil proceeding, a State is required by “substantive” due process to prove by clear and convincing evidence that the person is both (1) mentally ill, and (2) a danger to himself or to others. Id., at 259, 912 P. 2d, at 137. The court then determined that the Act’s definition of “mental abnormality” did not satisfy what it perceived to be this Court’s “mental illness” requirement in the civil commitment context. As a result, the court held that “the Act violates Hendricks’ substantive due process rights.” Id., at 261, 912 P. 2d, at 138.

The majority did not address Hendricks’ ex post facto or double jeopardy claims. The dissent, however, considered each of Hendricks’ constitutional arguments and rejected them. Id., at 264-294, 912 P. 2d, 140-156 (Larson, J., dissenting).

. II

A

Kansas argues that the Act’s definition of “mental abnormality” satisfies “substantive” due process requirements. We agree. Although freedom from physical restraint “has always been at the core of the liberty protected by the Due Process Clause from arbitrary governmental action,” Foucha v. Louisiana, 504 U. S. 71, 80 (1992), that liberty interest is not absolute. The Court has recognized that an individual’s constitutionally protected interest in avoiding physical restraint may be overridden even in the civil context:

“[T]he liberty secured by the Constitution of the United States to every person within its jurisdiction does not *357import an absolute right in each person to be, at all times and in all circumstances, wholly free from restraint. There are manifold restraints to which every person is necessarily subject for the common good. On any other basis organized society could not exist with safety to its members.” Jacobson v. Massachusetts197 U. S. 11, 26 (1905).

Accordingly, States have in certain narrow circumstances provided for the forcible civil detainment of people who are unable to control their behavior and who thereby pose a danger to the public health and safety. See, e. g., 1788 N. Y. Laws, ch. 31 (Feb. 9, 1788) (permitting confinement of the “furiously mad”); see also A. Deutsch, The Mentally Ill in America (1949) (tracing history of civil commitment in the 18th and 19th centuries); G. Grob, Mental Institutions in America: Social Policy to 1875 (1973) (discussing colonial and early American civil commitment statutes). We have consistently upheld such involuntary commitment statutes provided the confinement takes place pursuant to proper procedures and evidentiary standards. See Foucha, supra, at 80; Addington v. Texas, 441 U. S. 418, 426-427 (1979). It thus cannot be said that the involuntary civil confinement of a limited subclass of dangerous persons is contrary to our understanding of ordered liberty. Cf. id., at 426.

The challenged Act unambiguously requires a finding of dangerousness either to one’s self or to others as a prerequisite to involuntary confinement. Commitment proceedings can be initiated only when a person “has been convicted of or charged with a sexually violent offense,” and “suffers from a mental abnormality or personality disorder which makes the person likely to engage in the predatory acts of sexual violence.” Kan. Stat. Ann. § 59-29a02(a) (1994). The statute thus requires proof of more than a mere predisposition to violence; rather, it requires evidence of past sexually violent behavior and a present mental condition that creates a likelihood of such conduct in the future if the person is not inca*358pacitated. As we have recognized, “[previous instances of violent behavior are an important indicator of future violent tendencies.” Heller v. Doe, 509 U. S. 312, 323 (1993); see also Schall v. Martin, 467 U. S. 253, 278 (1984) (explaining that “from a legal point of view there is nothing inherently unattainable about a prediction of future criminal conduct”).

A finding of dangerousness, standing alone, is ordinarily not a sufficient ground upon which to justify indefinite involuntary commitment. We have sustained civil commitment statutes when they have coupled proof of dangerousness with the proof of some additional factor, such as a “mental illness” or “mental abnormality.” See, e. g., Heller, supra, at 314-315 (Kentucky statute permitting commitment of “mentally retarded” or “mentally ill” and dangerous individual); Allen v. Illinois, 478 U. S. 364, 366 (1986) (Illinois statute permitting commitment of “mentally ill” and dangerous individual); Minnesota ex rel. Pearson v. Probate Court of Ramsey Cty., 309 U. S. 270, 271-272 (1940) (Minnesota statute permitting commitment of dangerous individual with “psychopathic personality”). These added statutory requirements serve to limit involuntary civil confinement to those who suffer from a volitional impairment rendering them dangerous beyond their control. The Kansas Act is plainly of a kind with these other civil commitment statutes: It requires a finding of future dangerousness, and then links that finding to the existence of a “mental abnormality” or “personality disorder” that makes it difficult, if not impossible, for the person to control his dangerous behavior. Kan. Stat. Ann. § 59-29a02(b) (1994). The precommitment requirement of a “mental abnormality” or “personality disorder” is consistent with the requirements of these other statutes that we have upheld in that it narrows the class of persons eligible for confinement to those who are unable to control their dangerousness.

Hendricks nonetheless argues that our earlier cases dictate a finding of “mental illness” as a prerequisite for civil commitment, citing Foucha and Addington. He then as*359serts that a “mental abnormality” is not equivalent to a “mental illness” because it is a term coined by the Kansas Legislature, rather than by the psychiatric community. Contrary to Hendricks’ assertion, the term “mental illness” is devoid of any talismanic significance. Not only do “psychiatrists disagree widely and frequently on what constitutes mental illness,” Ake v. Oklahoma, 470 U. S. 68, 81 (1985), but the Court itself has used a variety of expressions to describe the mental condition of those properly subject to civil confinement. See, e. g., Addington, supra, at 425-426 (using the terms “emotionally disturbed” and “mentally ill”); Jackson v. Indiana, 406 U. S. 715, 732, 737 (1972) (using the terms “incompetency” and “insanity”); cf. Foucha, 504 U. S., at 88 (O’Connor, J., concurring in part and concurring in judgment) (acknowledging State’s authority to commit a person when there is “some medical justification for doing so”).

Indeed, we have never required state legislatures to adopt any particular nomenclature in drafting civil commitment statutes. Rather, we have traditionally left to legislators the task of defining terms of a medical nature that have legal significance. Cf. Jones v. United States, 463 U. S. 354, 365, n. 13 (1983). As a consequence, the States have, over the years, developed numerous specialized terms to define mental health concepts. Often, those definitions do not fit precisely with the definitions employed by the medical community. The legal definitions of “insanity” and “competency,” for example, vary substantially from their psychiatric counterparts. See, e. g., Gerard, The Usefulness of the Medical Model to the Legal System, 39 Rutgers L. Rev. 377, 391-394 (1987) (discussing differing purposes of legal system and the medical profession in recognizing mental illness). Legal definitions, however, which must “take into account such issues as individual responsibility . . . and competency,” need not mirror those advanced by the medical profession. American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders xxiii, xxvii (4th ed. 1994).

*360To the extent that the civil commitment statutes we have considered set forth criteria relating to an individual’s inability to control his dangerousness, the Kansas Act sets forth comparable criteria and Hendricks’ condition doubtless satisfies those criteria. The mental health professionals who evaluated Hendricks diagnosed him as suffering from pedophilia, a condition the psychiatric profession itself classifies as a serious mental disorder. See, e. g., id., at 524-525, 527-528; 1 American Psychiatric Association, Treatments of Psychiatric Disorders 617-633 (1989); Abel & Rouleau, Male Sex Offenders, in Handbook of Outpatient Treatment of Adults 271 (M. Thase, B. Edelstein, & M. Hersen eds. 1990).3 Hendricks even conceded that, when he becomes “stressed out,” he cannot “control the urge” to molest children. App. 172. This admitted lack of volitional control, coupled with a prediction of future dangerousness, adequately distinguishes Hendricks from other dangerous persons who are perhaps more properly dealt with exclusively through criminal proceedings. Hendricks’ diagnosis as a pedophile, which qualifies as a “mental abnormality” under the Act, thus plainly suffices for due process purposes.

B

We granted Hendricks’ cross-petition to determine whether the Act violates the Constitution’s double jeopardy *361prohibition or its ban on ex post facto lawmaking. The thrust of Hendricks’ argument is that the Act establishes criminal proceedings; hence confinement under it necessarily constitutes punishment. He contends that where, as here, newly enacted “punishment” is predicated upon past conduct for which he has already been convicted and forced to serve a prison sentence, the Constitution’s Double Jeopardy and Ex Post Facto Clauses are violated. We are unpersuaded by Hendricks’ argument that Kansas has established criminal proceedings.

The categorization of a particular proceeding as civil or criminal “is first of all a question of statutory construction.” Allen, 478 U. S., at 368. We must initially ascertain whether the legislature meant the statute to establish “civil” proceedings. If so, we ordinarily defer to the legislature’s stated intent. Here, Kansas’ objective to create a civil proceeding is evidenced by its placement of the Act within the Kansas probate code, Kan. Stat. Ann., Art. 29 (1994) (“Care and Treatment for Mentally Ill Persons”), instead of the criminal code, as well as its description of the Act as creating a “civil commitment procedure,” § 59-29a01 (emphasis added). Nothing on the face of the statute suggests that the legislature sought to create anything other than a civil commitment scheme designed to protect the public from harm.

Although we recognize that a “civil label is not always dis-positive,” Allen, supra, at 369, we will reject the legislature’s manifest intent only where a party challenging the statute provides “the clearest proof” that “the statutory scheme [is] so punitive either in purpose or effect as to negate [the State’s] intention” to deem it “civil,” United States v. Ward, 448 U. S. 242, 248-249 (1980). In those limited circumstances, we will consider the statute to have established criminal proceedings for constitutional purposes. Hendricks, however, has failed to satisfy this heavy burden.

As a threshold matter, commitment under the Act does not implicate either of the two primary objectives of criminal *362punishment: retribution or deterrence. The Act’s purpose is not retributive because it does not affix culpability for prior criminal conduct. Instead, such conduct is used solely for evidentiary purposes, either to demonstrate that a “mental abnormality” exists or to support a finding of future dangerousness. We have previously concluded that an Illinois statute was nonpunitive even though it was triggered by the commission of a sexual assault, explaining that evidence of the prior criminal conduct was “received not to punish past misdeeds, but primarily to show the accused’s mental condition and to predict future behavior.” Allen, supra, at 371. In addition, the Kansas Act does not make a criminal conviction a prerequisite for commitment — persons absolved of criminal responsibility may nonetheless be subject to confinement under the Act. See Kan. Stat. Ann. § 59-29a03(a) (1994). An absence, of the necessary criminal responsibility suggests that the State is not seeking retribution for a past misdeed. Thus, the fact that the Act may be “tied to criminal activity” is “insufficient to render the statut[e] punitive.” United States v. Ursery, 518 U. S. 267 (1996).

Moreover, unlike a criminal statute, no finding of scienter is required to commit an individual who is found to be a sexually violent predator; instead, the commitment determination is made based on a “mental abnormality” or “personality disorder” rather than on one’s criminal intent. The existence of a scienter requirement is customarily an important element in distinguishing criminal from civil statutes. See Kennedy v. Mendoza-Martinez, 372 U. S. 144, 168 (1963). The absence of such a requirement here is evidence that confinement under the statute is not intended to be retributive.

Nor can it be said that the legislature intended the Act to function as a deterrent. Those persons committed under the Act are, by definition, suffering from a “mental abnormality” or a “personality disorder” that prevents them from exercising adequate control over their behavior. Such persons are therefore unlikely to be deterred by the threat of *363confinement. And the conditions surrounding that confinement do not suggest a punitive purpose on the State’s part. The State has represented that an individual confined under the Act is not subject to the more restrictive conditions placed on state prisoners, but instead experiences essentially the same conditions as any involuntarily committed patient in the state mental institution. App. 50-56, 59-60. Because none of the parties argues that people institutionalized under the Kansas general civil commitment statute are subject to punitive conditions, even though they may be involuntarily confined, it is difficult to conclude that persons confined under this Act are being “punished.”

Although the civil commitment scheme at issue here does involve an affirmative restraint, “the mere fact that a person is detained does not inexorably lead to the conclusion that the government has imposed punishment.” United States v. Salerno, 481 U. S. 739, 746 (1987). The State may take measures to restrict the freedom of the dangerously mentally ill. This is a legitimate nonpunitive governmental objective and has been historically so regarded. Cf. id., at 747. The Court has, in fact, cited the confinement of “mentally unstable individuals who present a danger to the public” as one classic example of nonpunitive detention. Id., at 748-749. If detention for the purpose of protecting the community from harm necessarily constituted punishment, then all involuntary civil commitments would have to be considered punishment. But we have never so held.

Hendricks focuses on his confinement’s potentially indefinite duration as evidence of the State’s punitive intent. That focus, however, is misplaced. Far from any punitive objective, the confinement’s duration is instead linked to the stated purposes of the commitment, namely, to hold the person until his mental abnormality no longer causes him to be a threat to others. Cf. Jones, 463 U. S., at 368 (noting with approval that “because it is impossible to predict how long it will take for any given individual to recover [from insan*364ity] — or indeed whether he will ever recover — Congress has chosen ... to leave the length of commitment indeterminate, subject to periodic review of the patient’s suitability for release”). If, at any time, the confined person is adjudged “safe to be at large,” he is statutorily entitled to immediate release. Kan. Stat. Ann. § 59-29a07 (1994).

Furthermore, commitment under the Act is only 'potentially indefinite. The maximum amount of time an individual can be incapacitated pursuant to a single judicial proceeding is one year. §59-29a08. If Kansas seeks to continue the detention beyond that year, a court must once again determine beyond a reasonable doubt that the detainee satisfies the same standards as required for the initial confinement. Ibid. This requirement again demonstrates that Kansas does not intend an individual committed pursuant to the Act to remain confined any longer than he suffers from a mental abnormality rendering him unable to control his dangerousness.

' Hendricks next contends that the State’s use of procedural safeguards traditionally found in criminal trials makes the proceedings here criminal rather than civil. In Allen, we confronted a similar argument. There, the petitioner “place[d] great reliance on the fact that proceedings under the Act are accompanied by procedural safeguards usually found in criminal trials” to argue that the proceedings were civil in name only. 478 U. S., at 371. We rejected that argument, however, explaining that the State’s decision “to provide some of the safeguards applicable in criminal trials cannot itself turn these proceedings into criminal prosecutions.” Id., at 372. The numerous procedural and eviden-tiary protections afforded here demonstrate that the Kansas Legislature has taken great care to confine only a narrow class of particularly dangerous individuals, and then only after meeting the strictest procedural standards. That Kansas chose to afford such procedural protections does not *365transform a civil commitment proceeding into a criminal prosecution.

Finally, Hendricks argues that the Act is necessarily punitive because it fails to offer any legitimate “treatment.” Without such treatment, Hendricks asserts, confinement under the Act amounts to little more than disguised punishment. Hendricks’ argument assumes that treatment for his condition is available, but that the State has failed (or refused) to provide it. The Kansas Supreme Court, however, apparently rejected this assumption, explaining:

“It is clear that the overriding concern of the legislature is to continue the segregation of sexually violent offenders from the public. Treatment with the goal of reintegrating them into society is incidental, at best. The record reflects that treatment for sexually violent predators is all but nonexistent. The legislature concedes that sexually violent predators are not amenable to treatment under [the existing Kansas involuntary commitment statute]. If there is nothing to treat under [that statute], then there is no mental illness. In that light, the provisions of the Act for treatment appear somewhat disingenuous.” 259 Kan., at 258, 912 P. 2d, at 136.

It is possible to read this passage as a determination that Hendricks’ condition was untreatable under the existing Kansas civil commitment statute, and thus the Act’s sole purpose was incapacitation. Absent a treatable mental illness, the Kansas court concluded, Hendricks could not be detained against his will.

Accepting the Kansas court’s apparent determination that treatment is not possible for this category of individuals does not obligate us to adopt its legal conclusions. We have already observed that, under the appropriate circumstances and when accompanied by proper procedures, incapacitation *366may be a legitimate end of the civil law. See Allen, supra, at 373; Salerno, 481 U. S., at 748-749. Accordingly, the Kansas court’s determination that the Act’s “overriding concern” was the continued “segregation of sexually violent offenders” is consistent with our conclusion that the Act establishes civil proceedings, 259 Kan., at 258, 912 P. 2d, at 136, especially when that concern is coupled with the State’s ancillary goal of providing treatment to those offenders, if such is possible. While we have upheld state civil commitment statutes that aim both to incapacitate and to treat, see Allen, supra, we have never held that the Constitution prevents a State from civilly detaining those for whom no treatment is available, but who nevertheless pose a danger to others. A State could hardly be seen as furthering a “punitive” purpose by involuntarily confining persons afflicted with an untreatable, highly contagious disease. Accord, Compagnie Francaise de Navigation a Vapeur v. Louisiana Bd. of Health, 186 U. S. 380 (1902) (permitting involuntary quarantine of persons suffering from communicable diseases). Similarly, it would be of little value to require treatment as a precondition for civil confinement of the dangerously insane when no acceptable treatment existed. To conclude otherwise would obligate a State to release certain confined individuals who were both mentally ill and dangerous simply because they could not be successfully treated for their afflictions. Cf. Greenwood v. United States, 350 U. S. 366, 375 (1956) (“The fact that at present there may be little likelihood of recovery does not defeat federal power to make this initial commitment of the petitioner”); O’Connor v. Donaldson, 422 U. S. 563, 584 (1975) (Burger, C. J., concurring) (“[I]t remains a stubborn fact that there are many forms of mental illness which are not understood, some which are untreatable in the sense that no effective therapy has yet been discovered for them, and that rates of ‘cure’ are generally low”).

Alternatively, the Kansas Supreme Court’s opinion can be read to conclude that Hendricks’ condition is treatable, but *367that treatment was not the State’s “overriding concern,” and that no treatment was being provided (at least at the time Hendricks was committed). 259 Kan., at 258, 912 P. 2d, at 136. See also ibid. (“It is clear that the primary objective of the Act is to continue incarceration and not to provide treatment”). Even if we accept this determination that the provision of treatment was not the Kansas Legislature’s “overriding” or “primary” purpose in passing the Act, this does not rule out the possibility that an ancillary purpose of the Act was to provide treatment, and it does not require us to conclude that the Act is punitive. Indeed, critical language in the Act itself demonstrates that the Secretary, under whose custody sexually violent predators are committed, has an obligation to provide treatment to individuals like Hendricks. § 59-29a07(a) (“If the court or jury determines that the person is a sexually violent predator, the person shall be committed to the custody of the secretary of social and rehabilitation services for control, care and treatment until such time as the person’s mental abnormality or personality disorder has so changed that the person is safe to be at large” (emphasis added)). Other of the Act’s sections echo this obligation to provide treatment for committed persons. See, e. g., § 59-29a01 (establishing civil commitment procedure “for the long-term care and treatment of the sexually violent predator”); §59-29a09 (requiring the confinement to “conform to constitutional requirements for caré and treatment”). Thus, as in Allen, “the State has a statutory obligation to provide 'care and treatment for [persons adjudged sexually dangerous] designed to effect recovery,’ ” 478 U. S., at 369 (quoting Ill. Rev. Stat., ch. 38, ¶ 105-8 (1985)), and we may therefore conclude that “the State has ... provided for the treatment of those it commits,” 478 U. S., at 370.

Although the treatment program initially offered Hendricks may have seemed somewhat meager, it must be remembered that he was the first person committed under the *368Act. That the State did not have all of its treatment procedures in place is thus not surprising. What is significant, however, is that Hendricks was placed under the supervision of the Kansas Department of Health and Social and Rehabilitative Services, housed in a unit segregated from the general prison population and operated not by employees of the Department of Corrections, but by other trained individuals.4 And, before this Court, Kansas declared “ [absolutely” that persons committed under the Act are now receiving in the neighborhood of “31-1/2 hours of treatment per week.” Tr. of Oral Arg. 14-15,16.5

Where the State has “disavowed any punitive intent”; limited confinement to a small segment of particularly dangerous individuals; provided strict procedural safeguards; directed that confined persons be segregated from the general prison population and afforded the same status as others who have been civilly committed; recommended treatment if such is possible; and permitted immediate release upon a showing *369that the individual is no longer dangerous or mentally impaired, we cannot say that it acted with punitive intent. We therefore hold that the Act does not establish criminal proceedings and that involuntary confinement pursuant to the Act is not punitive. Our conclusion that the Act is nonpuni-tive thus removes an essential prerequisite for both Hendricks’ double jeopardy and ex post facto claims.

1

The Double Jeopardy Clause provides: “[N]or shall any person be subject for the same offence to be twice put in jeopardy of life or limb.” Although generally understood to preclude a second prosecution for the same offense, the Court has also interpreted this prohibition to prevent the State from “punishing twice, or attempting a second time to punish criminally, for the same offense.” Witte v. United States, 515 U. S. 389, 396 (1995) (emphasis and internal quotation marks omitted). Hendricks argues that, as applied to him, the Act violates double jeopardy principles because his confinement under the Act, imposed after a conviction and a term of incarceration, amounted to both a second prosecution and a second punishment for the same offense. We disagree.

Because we have determined that the Kansas Act is civil in nature, initiation of its commitment proceedings does not constitute a second prosecution. Cf. Jones v. United States, 463 U. S. 354 (1983) (permitting involuntary civil commitment after verdict of not guilty by reason of insanity). Moreover, as commitment under the Act is not tantamount to “punishment,” Hendricks’ involuntary detention does not violate the Double Jeopardy Clause, even though that confinement may follow a prison term. Indeed, in Baxstrom v. Herold, 383 U. S. 107 (1966), we expressly recognized that civil commitment could follow the expiration of a prison term without offending double jeopardy principles. We reasoned that “there is no conceivable basis for distinguishing the commitment of a person who is nearing the end of a penal *370term from all other civil commitments.” Id., at 111-112. If an individual otherwise meets the requirements for involuntary civil commitment, the State is under no obligation to release that individual simply because the detention would follow a period of incarceration.

Hendricks also argues that even if the Act survives the “multiple punishments” test, it nevertheless fails the “same elements” test of Blockburger v. United States, 284 U. S. 299 (1932). Under Blockburger, “where the same act or transaction constitutes a violation of two distinct statutory provisions, the test to be applied to determine whether there are two offenses or only one, is whether each provision requires proof of a fact which the other does not.” Id., at 304. The Blockburger test, however, simply does not apply outside of the successive prosecution context. A proceeding under the Act does not define an “offense,” the elements of which can be compared to the elements of an offense for which the person may previously have been convicted. Nor does the Act make the commission of a specified “offense” the basis for invoking the commitment proceedings. Instead, it uses a prior conviction (or previously charged conduct) for eviden-tiary purposes to determine whether a person suffers from a “mental abnormality” or “personality disorder” and also poses a threat to the public. Accordingly, we are unpersuaded by Hendricks’ novel application of the Blockburger test and conclude that the Act does not violate the Double Jeopardy Clause.

2

Hendricks’ ex post facto claim is similarly flawed. The Ex Post Facto Clause, which “ ‘forbids the application of any new punitive measure to a crime already consummated,’” has been interpreted to pertain exclusively to penal statutes. California Dept. of Corrections v. Morales, 514 U. S. 499, 505 (1995) (quoting Lindsey v. Washington, 301 U. S. 397, 401 (1937)). As we have previously determined, the Act does not impose punishment; thus, its application does not raise *371 ex post facto concerns. Moreover, the Act clearly does not have retroactive effect. Rather, the Act permits involuntary confinement based upon a determination that the person currently both suffers from a “mental abnormality” or “personality disorder” and is likely to pose a future danger to the public. To the extent that past behavior is taken into account, it is used, as noted above, solely for evidentiary purposes. Because the Act does not criminalize conduct legal before its enactment, nor deprive Hendricks of any defense that was available to him at the time of his crimes, the Act does not violate the Ex Post Facto Clause.

1 — 4

We hold that the Kansas Sexually Violent Predator Act comports with due process requirements and neither runs afoul of double jeopardy principles nor constitutes an exercise in impermissible ex post facto lawmaking. Accordingly, the judgment of the Kansas Supreme Court is reversed.

It is so ordered.

Justice Kennedy,

concurring.

I join the opinion of the Court in full- and add these additional comments.

Though other issues were argued to us, as the action has matured it turns on whether the Kansas statute is an ex post facto law. A law enacted after commission of the offense and which punishes the offense by extending the term of confinement is a textbook example of an ex post facto law. If the object or purpose of the Kansas law had been to provide treatment but the treatment provisions were adopted as a sham or mere pretext, there would have been an indication of the forbidden purpose to punish. The Court’s opinion gives a full and complete explanation why an ex post facto challenge based on this contention cannot succeed in the action before us. All this, however, concerns Hendricks alone. My brief, further comment is to caution against dangers in*372herent when a civil confinement law is used in conjunction with the criminal process, whether or not the law is given retroactive application.

It seems the dissent, too, would validate the Kansas statute as to persons who committed the crime after its enactment, and it might even validate the statute as to Hendricks, assuming a reasonable level of treatment. As all Members of the Court seem to agree, then, the power of the State to confine persons who, by reason of a mental disease or mental abnormality, constitute a real, continuing, and serious danger to society is well established. Addington v. Texas, 441 U. S. 418, 426-427 (1979). Confinement of such individuals is permitted even if it is pursuant to a statute enacted after the crime has been committed and the offender has begun serving, or has all but completed serving, a penal sentence, provided there is no object or purpose to punish. See Baxstrom v. Herold, 383 U. S. 107, 111-112 (1966). The Kansas law, with its attendant protections, including yearly review and review at any time at the instance of the person confined, is within this pattern and tradition of civil confinement. In this action, the mental abnormality — pedophilia—is at least described in the DSM-IV. American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders 524-525, 527-528 (4th ed. 1994).

Notwithstanding its civil attributes, the practical effect of the Kansas law may be to impose confinement for life. At this stage of medical knowledge, although future treatments cannot be predicted, psychiatrists or other professionals engaged in treating pedophilia may be reluctant to find measurable success in treatment even after a long period and may be unable to predict that no serious danger will come from release of the detainee.

A common response to this may be, “A life term is exactly what the sentence should have been anyway,” or, in the words of a Kansas task force member, “SO BE IT.” Testimony of Jim Blaufuss, App. 503. The point, however, is not *373how long Hendricks and others like him should serve a criminal sentence. With his criminal record, after all, a life term may well have been the only sentence appropriate to protect society and vindicate the wrong. The concern instead is whether it is the criminal system or the civil system which should make the decision in the first place. If the civil system is used simply to impose punishment after the State makes an improvident plea bargain on the criminal side, then it is not performing its proper function. These concerns persist whether the civil confinement statute is put on the books before or after the offense. We should bear in mind that while incapacitation is a goal common to both the criminal and civil systems of confinement, retribution and general deterrence are reserved for the criminal system alone.

On the record before us, the Kansas civil statute conforms to our precedents. If, however, civil confinement were to become a mechanism for retribution or general deterrence, or if it were shown that mental abnormality is too imprecise a category to offer a solid basis for concluding that civil detention is justified, our precedents would not suffice to validate it.

Justice Breyer,

with whom Justice Stevens and Justice Souter join, and with whom Justice Ginsburg joins as to Parts II and III, dissenting.

I agree with the majority that the Kansas Sexually Violent Predator Act’s “definition of ‘mental abnormality’ ” satisfies the “substantive” requirements of the Due Process Clause. Ante, at 356. Kansas, however, concedes that Hendricks’ condition is treatable; yet the Act did not provide Hendricks (or others like him) with any treatment until after his release date from prison and only inadequate treatment thereafter. These, and certain other, special features of the Act convince me that it was not simply an effort to commit Hendricks civilly, but rather an effort to inflict further punishment upon him. The Ex Post Facto Clause therefore pro*374hibits the Act’s application to Hendricks, who committed his crimes prior to its enactment.

I — (

I begin with the area of agreement. This Court has held that the civil commitment of a “mentally ill” and “dangerous” person does not automatically violate the Due Process Clause provided that the commitment takes place pursuant to proper procedures and evidentiary standards. See Foucha v. Louisiana, 504 U. S. 71, 80 (1992); Addington v. Texas, 441 U. S. 418, 426-427 (1979). The Kansas Supreme Court, however, held that the Due Process Clause forbids application of the Act to Hendricks for “substantive” reasons, i. e., irrespective of the procedures or evidentiary standards used. The court reasoned that Kansas had not satisfied the “mentally ill” requirement of the Due Process Clause because Hendricks was not “mentally ill.” In re Hendricks, 259 Kan. 246, 260-261, 912 P. 2d 129, 137-138 (1996). Moreover, Kansas had not satisfied what the court believed was an additional “substantive due process” requirement, namely, the provision of treatment. Id., at 257-258, 912 P. 2d, at 136. I shall consider each of these matters briefly.

A

In my view, the Due Process Clause permits Kansas to classify Hendricks as a mentally ill and dangerous person for civil commitment purposes. Allen v. Illinois, 478 U. S. 364, 370-371, 373-375 (1986). I agree with the majority that the Constitution gives States a degree of leeway in making this kind of determination. Ante, at 359; Foucha, supra, at 87 (O’Connor, J., concurring in part and concurring in judgment); Jones v. United States, 463 U. S. 354, 365, n. 13 (1983). But, because I do not subscribe to all of its reasoning, I shall set forth three sets of circumstances that, taken together, convince me that Kansas has acted within the limits that the Due Process Clause substantively sets.

*375First, the psychiatric profession itself classifies the kind of problem from whieh Hendricks suffers as a serious mental disorder. E. g., American Psychiatric Assn., Diagnostic and Statistical Manual of Mental Disorders 524-525, 527-528 (4th ed. 1994) (describing range of paraphilias and discussing how stress aggravates pedophilic behavior); Abel & Rouleau, Male Sex Offenders, in Handbook of Outpatient Treatment of Adults 271 (M. Thase, B. Edelstein, & M. Hersen eds. 1990). I concede that professionals also debate whether or not this disorder should be called a mental “illness.” See R. Slovenko, Psychiatry and Criminal Culpability 57 (1995) (citing testimony that paraphilias are not mental illnesses); Schopp & Sturgis, Sexual Predators and Legal Mental Illness for Civil Commitment, 13 Behav. Sci. & The Law 437, 451-452 (1995) (same). Compare Brief for American Psychiatric Association as Amicus Curiae 26 (mental illness requirement not satisfied) with Brief for Menninger Clinic et al. as Amici Curiae 22-25 (requirement is satisfied). But the very presence and vigor of this debate is important. The Constitution permits a State to follow one reasonable professional view, while rejecting another. See Addington v. Texas, supra, at 431. The psychiatric debate, therefore, helps to inform the law by setting the bounds of what is reasonable, but it cannot here decide just how States must write their laws within those bounds. See Jones, supra, at 365, n. 13.

Second, Hendricks’ abnormality does not consist simply of a long course of antisocial behavior, but rather it includes a specific, serious, and highly unusual inability to control his actions. (For example, Hendricks testified that, when he gets “stressed out,” he cannot “control the urge” to molest children, see ante, at 355.) The law traditionally has considered. this kind of abnormality, akin to insanity for purposes of confinement. See, e. g., Minnesota ex rel. Pearson v. Probate Court of Ramsey Cty., 309 U. S. 270, 274 (1940) (upholding against a due process challenge the civil confinement of *376a dangerous person where the danger flowed from an “ 'utter lack of power to control... sexual impulses’ ”) (quoting State ex rel. Pearson v. Probate Court of Ramsey Cty., 205 Minn. 545, 555, 287 N. W. 297, 302 (1939)); 1788 N. Y. Laws, ch. 31 (permitting confinement of those who are “furiously mad”); In re Oakes, 8 Law Rep. 122, 125 (Mass. 1845) (Shaw, C. J.); A. Deutsch, The Mentally Ill in America 419-420 (1949) (tracing history of commitment of furiously mad people in 18th and 19th centuries); Dershowitz, The Origins of Preventative Confinement in Anglo-American Law — Part II: The American Experience, 43 U. Cin. L. Rev. 781 (1974). Indeed, the notion of an “irresistible impulse” often has helped to shape criminal law’s insanity defense and to inform the related recommendations of legal experts as they seek to translate the insights of mental health professionals into workable legal rules. See also American Law Institute, Model Penal Code § 4.01 (insanity defense, in part, rests on inability “to conform . . . conduct to the requirements of law”); A. Gold-stein, The Insanity Defense 67-79 (1967) (describing “irresistible impulse” test).

Third, Hendricks’ mental abnormality also makes him dangerous. Hendricks “has been convicted of ... a sexually violent offense,” and a jury found that he “suffers from a mental abnormality ... which makes” him “likely to engage” in similar “acts of sexual violence” in the future. Kan. Stat. Ann. §§ 59-29a02, 59-29a03 (1994). The evidence at trial favored the State. Dr. Befort, for example, explained why Hendricks was likely to commit further acts of sexual violence if released. See, e. g., App. 248-254. And Hendricks’ own testimony about what happens when he gets “stressed out” confirmed Dr. Befort’s diagnosis.

Because (1) many mental health professionals consider pedophilia a serious mental disorder; and (2) Hendricks suffers from a classic case of irresistible impulse, namely, he is so afflicted with pedophilia that he cannot “control the urge” to molest children; and (3) his pedophilia presents a serious *377danger to those children, I believe that Kansas can classify Hendricks as “mentally ill” and “dangerous” as this Court used those terms in Foucha.

The Kansas Supreme Court’s contrary conclusion rested primarily upon that court’s view that Hendricks would not qualify for civil commitment under Kansas’ own state civil commitment statute. The issue before us, however, is one of constitutional interpretation. The Constitution does not require Kansas to write all of its civil commitment rules in a single statute or forbid it to write two separate statutes each covering somewhat different classes of committable individuals. Moreover, Hendricks apparently falls outside the scope of the Kansas general civil commitment statute because that statute permits confinement, only of those who “lac[k] capacity to make an informed decision concerning treatment.” Kan.,Stat. Ann. §59-2902(h) (1994). The statute does not tell us why it imposes this requirement. Capacity to make an informed decision about treatment is not always or obviously incompatible with severe mental illness. Neither Hendricks nor his amici point to a uniform body of professional opinion that says as much, and we have not found any. See, e.g., American Psychiatric Assn., Guidelines for Legislation on the Psychiatric Hospitalization of Adults, 140 Am. J. Psychiatry 672, 673 (1983); Stromberg & Stone, A Model State Law on Civil Commitment of the Mentally Ill, 20 Harv. J. Legis. 275, 301-302 (1983); DeLand & Borenstein, Medicine Court, II, Rivers in Practice, 147 Am. J. Psychiatry 38 (1990). Consequently, the boundaries of the Federal Constitution and those of Kansas’ general civil commitment statute are not congruent.

B

The Kansas Supreme Court also held that the Due Process Clause requires a State to provide treatment to those whom it civilly confines (as “mentally ill” and “dangerous”). It found that Kansas did not provide Hendricks with significant *378treatment. And it concluded that Hendricks’ confinement violated the Due Process Clause for this reason as well.

This case does not require us to consider whether the Due Process Clause always requires treatment — whether, for example, it would forbid civil confinement of an untreatable mentally ill, dangerous person. To the contrary, Kansas argues that pedophilia is an “abnormality” or “illness” that can be treated. See Tr. of Oral Arg. 12 (Kansas Attorney General, in response to the question “you’re claiming that there is some treatability . . . ?” answering “[absolutely”); Brief for Petitioner 42-47. Two groups of mental health professionals agree. Brief for Association for the Treatment of Sexual Abusers as Amicus Curiae 11-12 (stating that “sex offenders can be treated” and that “increasing evidence” shows that “state-of-the-art treatment programs . . . significantly reduce recidivism”); Brief for Menninger Foundation et al. as Amici Curiae 28. Indeed, no one argues the contrary. Hence the legal question before us is whether the Clause forbids Hendricks’ confinement unless Kansas provides him with treatment that it concedes is available.

Nor does anyone argue that Kansas somehow could have violated the Due Process Clause’s treatment concerns had it provided Hendricks with the treatment that is potentially available (and I do not see how any such argument could succeed). Rather, the basic substantive due process treatment question is whether that Clause requires Kansas to provide treatment that it concedes is potentially available to a person whom it concedes is treatable. This same question is at the heart of my discussion of whether Hendricks’ confinement violates the Constitution’s Ex Post Facto Clause. See infra, at 383-395. For that reason, I shall not consider the substantive due process treatment question separately, but instead shall simply turn to the Ex Post Facto Clause discussion. As Justice Kennedy points out, ante, p. 371, some of the matters there discussed may later prove relevant to substantive due process analysis.

*379h — i HH

Kansas’ 1994 Act violates the Federal Constitution’s prohibition of “any ... ex post facto Law” if it “inflicts” upon Hendricks “a greater punishment” than did the law “annexed to” his “erime[s]” when he “committed” those crimes in 1984. Colder v. Bull, 3 Dall. 386, 390 (1798) (opinion of Chase, J.); U. S. Const., Art. I, § 10. The majority agrees that the Clause “ ‘forbids the application of any new punitive measure to a crime already consummated.’” California Dept. of Corrections v. Morales, 514 U. S. 499, 505 (1995) (citation omitted; emphasis added). Ante, at 370-371. But it finds the Act is not “punitive.” With respect to that basic question, I disagree with the majority.

Certain resemblances between the Act’s “civil commitment” and traditional criminal punishments are obvious. Like criminal imprisonment, the Act’s civil commitment amounts to “secure” confinement, Kan. Stat. Ann. § 59— 29a07(a) (1994), and “incarceration against one’s will,” In re Gault, 387 U. S. 1, 50 (1967). See Testimony of Terry Davis, SRS Director of Quality Assurance, App. 52-54, 78-81 (confinement takes place in the psychiatric wing of a prison hospital where those whom the Act confines and ordinary prisoners are treated alike). Cf. Browning-Ferris Industries of Vt., Inc. v. Kelco Disposal, Inc., 492 U. S. 257, 298 (1989) (O’Connor, J., concurring in part and dissenting in part). In addition, a basic objective of the Act is incapacitation, which, as Blaekstone said in describing an objective of criminal law, is to “depriv[e] the party injuring of the power to do future mischief.” 4 W. Blaekstone, Commentaries *11-* 12 (incapacitation is one important purpose of criminal punishment); see also Foucha, 504 U. S., at 99 (Kennedy, J., dissenting) (“Incapacitation for the protection of society is not an unusual ground for incarceration”); United States v. Brown, 381 U. S. 437, 458 (1965) (“Punishment serves several purposes: retributive, rehabilitative, deterrent — and preventative. One of the reasons society imprisons those convicted *380of crimes is to keep them from inflicting future harm, but that does not make imprisonment any the less punishment”); 1 W. LaFave & A. Scott, Substantive Criminal Law § 1.5, p. 32 (1986); 18 U. S. C. § 3553(a); United States Sentencing Guidelines, Guidelines Manual, ch. 1, pt. A (Nov. 1995).

Moreover, the Act, like criminal punishment, imposes its confinement (or sanction) only upon an individual who has previously committed a criminal offense. Kan. Stat. Ann. §§ 59-29a02(a), 59-29a03(a) (1994). Cf. Department of Revenue of Mont. v. Kurth Ranch, 511 U. S. 767, 781 (1994) (fact that a tax on marijuana was “conditioned on the commission of a crime” is “ ‘significant of [its] penal and prohibitory intent’ ” (citation omitted)); Lipke v. Lederer, 259 U. S. 557, 561-562 (1922). And the Act imposes that confinement through the use of persons (county prosecutors), procedural guarantees (trial by jury, assistance of counsel, psychiatric evaluations), and standards (“beyond a reasonable doubt”) traditionally associated with the criminal law. Kan. Stat. Ann. §§ 59-29a06, 59-29a07 (1994).

These obvious resemblances by themselves, however, are not legally sufficient to transform what the Act calls “civil commitment” into a criminal punishment. Civil commitment of dangerous, mentally ill individuals by its very nature involves confinement and incapacitation. Yet “civil commitment,” from a constitutional perspective, nonetheless remains civil. Allen v. Illinois, 478 U. S., at 369-370. Nor does the fact that criminal behavior triggers the Act make the critical difference. The Act’s insistence upon a prior crime, by screening out those whose past behavior does not concretely demonstrate the existence of a mental problem or potential future danger, may serve an important noncriminal evidentiary purpose. Neither is the presence of criminal law-type procedures determinative. Those procedures can serve an important purpose that in this context one might consider noncriminal, namely, helping to prevent judgmental *381mistakes that would wrongly deprive a person of important liberty. Id., at 371-372.

If these obvious similarities cannot by themselves prove that Kansas’ “civil commitment” statute is criminal, neither can the word “civil” written into the statute, §59-29a01, by itself prove the contrary. This Court has said that only the “clearest proof” could establish that a law the legislature called “civil” was, in reality, a “punitive” measure. United States v. Ward, 448 U. S. 242, 248-249 (1980). But the Court has also reiterated that a “civil label is not always disposi-tive,” Allen v. Illinois, supra, at 369; it has said that in close cases the label is “ ‘not of paramount importance,’ ” Kurth Ranch, supra, at 777 (citation omitted); and it has looked behind a “civil” label fairly often, e. g., United States v. Halper, 490 U. S. 435, 447 (1989).

In this circumstance, with important features of the Act pointing in opposite directions, I would place particular importance upon those features that would likely distinguish between a basically punitive and a basically nonpunitive purpose. United States v. Ursery, 518 U. S. 267, 278 (1996) (asking whether a statutory scheme was so punitive “ ‘either in purpose or effect’ ” to negate the legislature’s “ ‘intention to establish a civil remedial mechanism’ ” (citations omitted)). And I note that the Court, in an earlier civil commitment case, Allen v. Illinois, 478 U. S., at 369, looked primarily to the law’s concern for treatment as an important distinguishing feature. I do not believe that Allen means that a particular law’s lack of concern for treatment, by itself, is enough to make an incapacitative law punitive. But, for reasons I will point out, when a State believes that treatment does exist, and then couples that admission with a legislatively required delay of such treatment until a person is at the end of his jail term (so that further incapacitation is therefore necessary), such a legislative scheme begins to look punitive.

In Allen, the Court considered whether, for Fifth Amendment purposes, proceedings under an Illinois statute were *382civil or “criminal.” The Illinois statute, rather like the Kansas statute here, authorized the confinement of persons who were sexually dangerous, who had committed at least one prior sexual assault, and who suffered from a “mental disorder.” Id., at 366, n. 1. The Allen Court, looking behind the statute’s “civil commitment” label, found the statute civil— in important part because the State had “provided for the treatment of those it commits.” Id., at 370 (also referring to facts that the State had “disavowed any interest in punishment” and that it had “established a system under which committed persons may be released after the briefest time in confinement”).

In reaching this conclusion, the Court noted that the State Supreme Court had found the proceedings “‘essentially civil’ ” because the statute’s aim was to provide “ ‘treatment, not punishment.’ ” Id., at 367 (quoting People v. Allen, 107 Ill. 2d 91, 99-101, 481 N. E. 2d 690, 694-695 (1985)). It observed that the State had “a statutory obligation to provide ‘care and treatment . . . designed to effect recovery’ ” in a “facility set aside to provide psychiatric care.” 478 U. S., at 369 (quoting Ill. Rev. Stat., ch. 38, ¶ 105-8 (1985)). And it referred to the State’s purpose as one of “treating rather than punishing sexually dangerous persons.” 478 U. S., at 373; see also ibid. (“Had petitioner shown, for example, that the confinement . . . imposes ... a regimen which is essentially identical to that imposed upon felons with no need for psychiatric care, this might well be a different case”).

The Allen Court’s focus upon treatment, as a kind of touchstone helping to distinguish civil from punitive purposes, is not surprising, for one would expect a nonpunitive statutory scheme to confine, not simply in order to protect, but also in order to cure. That is to say, one would expect a nonpunitively motivated legislature that confines because of a dangerous mental abnormality to seek to help the individual himself overcome that abnormality (at least insofar as professional treatment for the abnormality exists and is *383potentially helpful, as Kansas, supported by some groups of mental health professionals, argues is the case here, see supra, at 378). Conversely, a statutory scheme that provides confinement that does not reasonably fit a practically available, medically oriented treatment objective, more likely reflects a primarily punitive legislative purpose.

Several important treatment-related factors — factors of a kind that led the five-Member Allen majority to conclude that the Illinois Legislature’s purpose was primarily civil, not punitive — in this action suggest precisely the opposite. First, the State Supreme Court here, unlike the state court in Allen, has held that treatment is not a significant objective of the Act. The Kansas court wrote that the Act’s purpose is “segregation of sexually violent offenders,” with “treatment” a matter that was “incidental at best.” 259 Kan., at 258, 912 P. 2d, at 136. By way of contrast, in Allen the Illinois court had written that “‘treatment, not punishment,’” was “the aim of the statute.” Allen, supra, at 367 (quoting People v. Allen, supra, at 99-101, 481 N. E. 2d, at 694-695).

We have generally given considerable weight to the findings of state and lower federal courts regarding the intent or purpose underlying state officials’ actions, see U. S. Term Limits, Inc. v. Thornton, 514 U. S. 779, 829 (1995) (ordinarily “[w]e must . . . accept the state court’s view of the purpose of its own law”); Romer v. Evans, 517 U. S. 620, 626 (1996); Hernandez v. New York, 500 U. S. 352, 366-370 (1991) (plurality opinion); id., at 372 (O’Connor, J., concurring); Edwards v. Aguillard, 482 U. S. 578, 594, n. 15 (1987); b.ut see Department of Revenue of Mont. v. Kurth Ranch, 511 U. S., at 776, 780, n. 18; Stone v. Graham, 449 U. S. 39,40-43 (1980) (per curiam); Consolidated Edison Co. of N. Y. v. Public Serv. Comm’n of N. Y, 447 U. S. 530, 533, 535-537 (1980), although the level of deference given to such findings varies with the circumstances, Crawford v. Board of Ed. of Los Angeles, 458 U. S. 527, 544, n. 30 (1982), and is not always as conclusive as a state court’s construction of one of its stat*384utes, see, e. g., R. A. V. v. St. Paul, 505 U. S. 377, 381 (1992). For example, Allen’s dissenters, as well as its majority, considered the state court’s characterization of the state law’s purpose an important factor in determining the constitutionality of that statute. Allen, 478 U. S., at 380 (Stevens, J., dissenting) (describing the state court as “the final authority on the . .. purpose” of the statute).

The record provides support for the Kansas court’s conclusion. The court found that, as of the time of Hendricks’ commitment, the State had not funded treatment, it had not entered into treatment contracts, and it had little, if any, qualified treatment staff. See 259 Kan., at 249, 258, 912 P. 2d, at 131, 136; Testimony of Dr. Charles Befort, App. 255 (acknowledging that he has no specialized training); Testimony of John House, SRS Attorney, id., at 367 (no contract has been signed by bidders); Testimony of John House, SRS Attorney, id., at 369 (no one hired to operate sexually violent predator (SVP) program or to serve as clinical director, psychiatrist, or psychologist). Indeed, were we to follow the majority’s invitation to look beyond the record in this case, an invitation with which we disagree, see infra, at 391-393, it would reveal that Hendricks, according to the commitment program’s own director, was receiving “essentially no treatment.” Dr. Charles Befort in State Habeas Corpus Proceeding, App. 393; 259 Kan., at 249, 258, 912 P. 2d, at 131, 136. See also App. 421 (“[T]he treatment that is prescribed by statute” is “still not available”); id., at 420-421 (the “needed treatment” “hasn’t been delivered yet” and “Hendricks has wasted ten months” in “terms of treatment effects”); id., at 391-392 (Dr. Befort admitting that he is not qualified to be SVP program director).

It is therefore not surprising that some of the Act’s official supporters had seen in it an opportunity permanently to confine dangerous sex offenders, e. g., id., at 468 (statement of Attorney General Robert Stephan); id., at 475-476, 478 (statement of Special Assistant to the Attorney General *385Carla Stovall). Others thought that effective treatment did not exist, id., at 503 (statement of Jim Blaufuss) (“Because there is no effective treatment for sex offenders, this Bill may mean a life sentence for a felon that is considered a risk to women and children. SO BE IT!”) — a view, by the way, that the State of Kansas, supported by groups of informed mental health professionals, here strongly denies. See supra, at 378.

The Kansas court acknowledged the existence of “provisions of the Act for treatment” (although it called them “somewhat disingenuous”). 259 Kan., at 258, 912 P. 2d, at 136. Cf. Kan. Stat. Ann. § 59-29a01 (1994) (legislative findings that “prognosis for rehabilitation]... in a prison setting is poor,... treatment needs ... long term” and “commitment procedure for . . . long term care and treatment . . . necessary”); § 59-29a09 (“commitment... shall conform to constitutional requirements for care and treatment”). Nor did the court deny that Kansas could later increase the amount of treatment it provided. But the Kansas Supreme Court could, and did, use the Act’s language, history, and initial implementation to help it characterize the Act’s primary purposes.

Second, the Kansas statute, insofar as it applies to previously convicted offenders such as Hendricks, commits, confines, and treats those offenders after they have served virtually their entire criminal sentence. That time-related circumstance seems deliberate. The Act explicitly defers diagnosis, evaluation, and commitment proceedings until a few weeks prior to the “anticipated release” of a previously convicted offender from prison. Kan. Stat. Ann. §59-29a03(a)(1) (1994). But why, one might ask, does the Act not commit and require treatment of sex offenders sooner, say, soon after they begin to serve their sentences?

An Act that simply seeks confinement, of course, would not need to begin civil commitment proceedings sooner. Such an Act would have to begin proceedings only when an *386offender’s prison term ends, threatening his release from the confinement that imprisonment assures. But it is difficult to see why rational legislators who seek treatment would write the Act in this way — providing treatment years after the criminal act that indicated its necessity. See, e. g., Wettstein, A Psychiatric Perspective on Washington’s Sexually Violent Predators Statute, 15 U. Puget Sound L. Rev. 597, 617 (1992) (stating that treatment delay leads to “loss of memory” and makes it “more difficult for the offender” to “accept responsibility,” and that time in prison leads to attitude hardening that “engender[s] a distorted view of the precipitating offense”). And it is particularly difficult to see why legislators who specifically wrote into the statute a finding that “prognosis for rehabilitating ... in a prison setting is poor” would leave an offender in that setting for months or years before beginning treatment. This is to say, the timing provisions of the statute confirm the Kansas Supreme Court’s view that treatment was not a particularly important legislative objective.

I recognize one possible counterargument. A State, wanting both to punish Hendricks (say, for deterrence purposes) and also to treat him, might argue that it should be permitted to postpone treatment until after punishment in order to make certain that the punishment in fact occurs. But any such reasoning is out of place here. Much of the treatment that Kansas offered here (called “ward milieu” and “group therapy”) can be given at the same time as, and in the same place where, Hendricks serves his punishment. See, e. g., Testimony of Leroy Hendricks, App. 142-143, 150, 154, 179-181 (stating that Washington and Kansas had both provided group therapy to Hendricks, and that he had both taken and refused such treatment at various points); Testimony of Terry Davis, SRS Director of Quality Assurance, id., at 78-81 (pointing out that treatment under the Act takes place in surroundings very similar to those in which prisoners receive treatment); Testimony of John House, SRS *387Attorney, id., at 375-376. See also Task Force on Community Protection, Final Report to Booth Gardner, Governor State of Washington II—2 (1989) (findings of task force that developed the Washington State Act, which served as a model for Kansas’ Act, stating that “[s]ex offenders can be treated during incarceration”). The evidence adduced at the state habeas proceeding, were we to assume it properly before the Court, see infra, at 392-393, supports this conclusion as well. See Testimony of Dr. Befort at State Habeas Proceeding, App. 399, 406-408 (describing treatment as ward milieu and group therapy); id., at 416-417 (stating that Kansas offers similar treatment, on a voluntary basis, to prisoners). Hence, assuming, arguendo, that it would be otherwise permissible, Kansas need not postpone treatment in order to make certain that sex offenders serve their full terms of imprisonment, i. e., to make certain that they receive the entire punishment that Kansas criminal law provides. To the contrary, the statement in the Act itself, that the Act aims to respond to special “long term” “treatment needs,” suggests that treatment should begin during imprisonment. It also suggests that, were those long-term treatment needs (rather than further punishment) Kansas’ primary aim, the State would require that treatment begin soon after conviction, not 10 or more years later. See also Vt. Stat. Ann., Tit. 18, § 2815 (1959) (providing for treatment of sexual psychopaths first, and punishment afterwards).

Third, the statute, at least as of the time Kansas applied it to Hendricks, did not require the committing authority to consider the possibility of using less restrictive alternatives, such as postrelease supervision, halfway houses, or other methods that amici supporting Kansas here have mentioned. Brief for Menninger Foundation et al. as Amici Curiae 28; Brief for Association for the Treatment of Sexual Abusers as Amicus Curiae 11-12. The laws of many other States require such consideration. See Appendix, infra.

*388This Court has said that a failure to consider, or to use, “alternative and less harsh methods” to achieve a nonpuni-tive objective can help to show that legislature’s “purpose ... was to punish.” Bell v. Wolfish, 441 U. S. 520, 539, n. 20 (1979). And one can draw a similar conclusion here. Legislation that seeks to help the individual offender as well as to protect the public would avoid significantly greater restriction of an individual’s liberty than public safety requires. See Keilitz, Conn, & Gianpetro, Least Restrictive Treatment of Involuntary Patients: Translating Concepts into Practice, 29 St. Louis U. L. J. 691, 693 (1985) (describing “least restrictive alternativ[e]” provisions in the ordinary civil commitment laws of almost all States); Lyon, Levine, & Zusman, Patients’ Bill of Rights: A Survey of State Statutes, 6 Mental Disability L. Rep. 178, 181-183 (1982) (same). Legislation that seeks almost exclusively to incapacitate the individual through confinement, however, would not necessarily concern itself with potentially less restrictive forms of incapacitation. I would reemphasize that this is not a case in which the State claims there is no treatment potentially available. Rather, Kansas, and supporting amici, argue that pedophilia is treatable. See supra, at 378.

Fourth, the laws of other States confirm, through comparison, that Kansas’ “civil commitment” objectives do not require the statutory features that indicate a punitive purpose. I have found 17 States with laws that seek to protect the public from mentally abnormal, sexually dangerous individuals through civil commitment or other mandatory treatment programs. Ten of those statutes, unlike the Kansas statute, begin treatment of an offender soon after he has been apprehended and charged with a serious sex offense. Only seven, like Kansas, delay “civil” commitment (and treatment) until the offender has served his criminal sentence (and this figure includes the Acts of Minnesota and New Jersey, both of which generally do not delay treatment). Of these seven, however, six (unlike Kansas) require consideration of less re*389strictive alternatives. See Ariz. Rev. Stat. Ann. §§ 13-4601, 4606B (Supp. 1996-1997); Cal. Welf. & Inst. Code Ann. §§ 6607, 6608 (West Supp. 1997); Minn. Stat. §253B.09 (1996); N. J. Stat. Ann. §30:4-27.11d (West 1997); Wash. Rev. Code Ann. §71.09.090 (Supp. 1996-1997); Wis. Stat. § 980.06(2)(b) (Supp. 1993-1994). Only one State other than Kansas, namely Iowa, both delays civil commitment (and consequent treatment) and does not explicitly consider less restrictive alternatives. But the law of that State applies prospectively only, thereby avoiding ex post facto problems. See Iowa Code Ann. §709C.12 (Supp. 1997) (Iowa SVP Act only “applies to persons convicted of a sexually violent offense on or after July 1, 1997”); see also Appendix, infra. Thus the practical experience of other States, as revealed by their statutes, confirms what the Kansas Supreme Court’s finding, the timing of the civil commitment proceeding, and the failure to consider less restrictive alternatives, themselves suggest, namely, that for Ex Post Facto Clause purposes, the purpose of the Kansas Act (as applied to previously convicted offenders) has a punitive, rather than a purely civil, purpose.

Kansas points to several cases as support for a contrary conclusion. It points to Allen — which is, as we have seen, a ease in which the Court concluded that Illinois’ “civil commitment” proceedings were not criminal. I have explained in detail, however, how the statute here differs from that in Allen, and why Allen’s reasoning leads to a different conclusion in this litigation. See supra, at 381-388 and this page.

Kansas also points to Addington v. Texas, where the Court held that the Constitution does not require application of criminal law’s “beyond a reasonable doubt” standard in a civil commitment proceeding. 441 U. S., at 428. If some criminal law guarantees such as “reasonable doubt” did not apply in Addington, should other guarantees, such as the prohibition against ex post facto laws, apply here? The answer to this question, of course, lies in the particular statute at issue in Addington — a Texas statute that, this Court ob*390served, did “not exercis[e]” state power “in a punitive sense." Ibid. That statute did not add civil commitment’s confinement to imprisonment; rather civil commitment was, at most, a substitute for criminal punishment. See Tex. Rev. Civ. Stat. Ann. § 5547-41 (Vernon 1958) (petition must state “proposed patient is not charged with a crime or [is] charged [but] transferred ... for civil commitment proceedings”). And this Court, relying on the Texas Supreme Court’s interpretation, wrote that the “State of Texas confines only for the purpose of providing care designed to treat the individual." Addington, supra, at 428, n. 4 (citing State v. Turner, 556 S. W. 2d 563, 566 (1977)). Cf. Specht v. Patterson, 386 U. S. 605, 608-609 (1967) (separate postconviction sexual psychopath commitment/sentencing proceeding held after conviction for serious sex crime, imposes a “criminal punishment even though . . . designed not so much as retribution as ... to keep individuals from inflicting future harm”). Nothing I say here would change the reach or holding of Addington in any way. That, is, a State is free to commit those who are dangerous and mentally ill in order to treat them. Nor does my decision preclude a State from deciding that a certain subset of people are mentally ill, dangerous, and untreatable, and that confinement of this subset is therefore necessary (again, assuming that all the procedural safeguards of Addington are in place). But when a State decides offenders can be treated and confines an offender to provide that treatment, but then refuses to provide it, the refusal to treat while a person is fully incapacitated begins to look punitive.

The majority suggests that this is the very case I say it is not, namely, a case of a mentally ill person who is untreatable. Ante, at 365. And it quotes a long excerpt from the Kansas Supreme Court’s opinion in support. That court, however, did not find that Hendricks was untreatable; it found that he was untreated — quite a different matter. Had the Kansas Supreme Court thought that Hendricks, or oth*391ers like him, are untreatable, it could not have written the words that follow that excerpt, adopting by reference the words of another court opinion:

‘“The statute forecloses the possibility that offenders will be evaluated and treated until after they have been punished. . . . Setting aside the question of whether a prison term exacerbates or minimizes the mental condition of a sex offender, it plainly delays the treatment that must constitutionally accompany commitment pursuant to the Statute. The failure of the Statute to provide for examination or treatment prior to the completion of the punishment phase strongly suggests that treatment is of secondary, rather than primary, concern.’ ” 259 Kan., at 258, 912 P. 2d, at 136 (quoting Young v. Weston, 898 F. Supp. 744, 753 (WD Wash. 1995)).

This quotation, and the rest of the opinion, make clear that the court is finding it objectionable that the statute, among other things, has not provided adequate treatment to one who, all parties here concede, can be treated.

The majority suggests in the alternative that recent evidence shows that Kansas is now providing treatment. Ante, at 366-368. That evidence comes from two sources: First, a statement by the Kansas Attorney General at oral argument that those committed under the Act are now receiving treatment, ante, at 368; and second, in a footnote, a Kansas trial judge’s statement, in a state habeas proceeding nearly one year after Hendricks was committed, that Kansas is providing treatment. Ante, at 368, n. 5. I do not see how either of these statements can be used to justify the validity of the Act’s application to Hendricks at the time he filed suit.

We are reviewing the Kansas Supreme Court’s determination. Of Hendricks’ case. Neither the majority nor the lengthy dissent in that court referred to the two facts that the majority now seizes upon, and for good reason. That court denied a motion to take judicial notice of the state *392habeas proceeding, see Order of Kansas Supreme Court, No. 94-73039, Mar. 1,1996. The proceeding is thus not part of the record, and cannot properly be considered by this Court. And the Kansas Supreme Court obviously had no chance to consider Kansas’ new claim made at. oral argument before this Court. There is simply no evidence in the record before this Court that comes even close to resembling the assertion Kansas made at oral argument. It is the record, not the parties’ view of it, that must control our decision. See Russell v. Southard, 12 How. 139, 158-159 (1851); Adickes v. S. H. Kress & Co., 398 U. S. 144, 157-158, n. 16 (1970); Hopt v. Utah, 114 U. S. 488, 491-492 (1885); Witters v. Washington Dept. of Servs. for Blind, 474 U. S. 481, 489, n. 3 (1986); New Haven Inclusion Cases, 399 U. S. 392, 450, n. 66 (1970); R. Stern, E. Gressman, S. Shapiro, & K. Geller, Supreme Court Practice 555-556, 594 (7th ed. 1993); Fed. Rule Evid. 201(b).

The prohibition on facts found outside the record is designed to ensure the reliability of the evidence before the Court. For purposes of my argument in this dissent, however, the material that the majority wishes to consider, when read in its entirety, shows that Kansas was not providing treatment to Hendricks. At best, the testimony at the state hearing contained general and vague references that treatment was about to be provided, but it contains no statement that Hendricks himself was receiving treatment. And it provides the majority with no support at all in respect to that key fact. Indeed, it demonstrates the contrary conclusion. For example, the program’s director, Dr. Befort, testified that he would have to tell the court at Hendricks’ next annual review, in October 1995, that Hendricks “has had no opportunity for meaningful treatment.” App. 400. He also stated that SVP’s were receiving “essentially no treatment” and that the program does not “have adequate staffing.” Id., at 393,394. And Dr. Befort’s last words made clear that Hendricks has “wasted ten months ... in terms of treatment *393effects” and that, as far as treatment goes, “[t]oday, it’s still not available.” Id., at 420-421. Nor does the assertion made by the Kansas Attorney General at oral argument help the majority. She never stated that Hendricks, as opposed to other SVP’s, was receiving this treatment. And we can find no support for her statement in the record.

We have found no other evidence in the record to support the conclusion that Kansas was in fact providing the treatment that all parties agree that it could provide. Thus, even had the Kansas Supreme Court considered the majority’s new evidence — which it did not — it is not likely to have changed its characterization of the Act’s treatment provisions as “somewhat disingenuous.” 259 Kan., at 258, 912 P. 2d, at 186.

Regardless, the Kansas Supreme Court did so characterize the Act’s treatment provisions and did find that treatment was “at best” an “incidental” objective. Thus, the circumstances here are different from Allen, where the Illinois Supreme Court explicitly found that the statute’s aim was to provide treatment, not punishment. See supra, at 382-384. There is no evidence in the record that contradicts the finding of the Kansas court. Thus, Allen’s approach — its reliance on the state court — if followed here would mean the Act as applied to Leroy Hendricks (as opposed to others who may have received treatment or who were sentenced after the effective date of the Act) is punitive.

Finally, Kansas points to United States v. Salerno, 481 U. S. 739 (1987), a case in which this Court held preventive detention of a dangerous accused person pending trial constitutionally permissible. Salerno, however, involved the brief detention of that person, after a finding of “probable cause” that he had committed a crime that would justify further imprisonment, and only pending a speedy judicial determination of guilt or innocence. This Court, in Foucha, emphasized the fact that the confinement at issue in Salerno was “strictly limited in duration.” 504 U. S., at 82. It described *394that “pretrial detention of arrestees” as “one of those carefully limited exceptions permitted by the Due Process Clause.” Id., at 83. And it held that Salerno did not authorize the indefinite detention, on grounds of dangerousness, of “insanity acquittees who are not mentally ill but who do not prove they would not be dangerous to others.” 504 U. S., at 83. Whatever Salerno's “due process” implications may be, it does not focus upon, nor control, the question at issue here, the question of “punishment” for purposes of the Ex Post Facto Clause.

One other case warrants mention. In Kennedy v. Mendoza-Martinez, 372 U. S. 144 (1963), this Court listed seven factors that helped it determine whether a particular statute was primarily punitive for purposes of applying the Fifth and Sixth Amendments. Those factors include whether a sanction involves an affirmative restraint, how history has regarded it, whether it applies to behavior already a crime, the need for a finding of scienter, its relationship to a traditional aim of punishment, the presence of a nonpunitive alternative purpose, and whether it is excessive in relation to that purpose. Id., at 169. This Court has said that these seven factors are “neither exhaustive nor disposi-tive,” but nonetheless “helpful.” Ward, 448 U. S., at 249. Paraphrasing them here, I believe the Act before us involves an affirmative restraint historically regarded as punishment; imposed upon behavior already a crime after a finding of scienter; which restraint, namely, confinement, serves a traditional aim of punishment, does not primarily serve an alternative purpose (such as treatment), and is excessive in relation to any alternative purpose assigned. 372 U. S., at 168-169.

This is not to say that each of the factors the Court mentioned in Martinez-Mendoza on balance argues here in favor of a constitutional characterization as “punishment.” It is not to say that I have found “a single ‘formula’ for identifying those legislative changes that have a sufficient effect on *395substantive crimes or punishments to fall within the constitutional prohibition,” Morales, 514 U. S., at 509; see also Halper, 490 U. S., at 447; id., at 453 (Kennedy, J., concurring). We have not previously done so, and I do not do so here. Rather, I have pointed to those features of the Act itself, in the context of this litigation, that lead me to conclude, in light of our precedent, that the added confinement the Act imposes upon Hendricks is basically punitive. This analysis, rooted in the facts surrounding Kansas’ failure to treat Hendricks, cannot answer the question whether the Kansas Act, as it now stands, and in light of its current implementation, is punitive toward people other than he. And I do not attempt to do so here.

Ill

To find that the confinement the Act imposes upon Hendricks is “punishment” is to find a violation of the Ex Post Facto Clause. Kansas does not deny that the 1994 Act changed the legal consequences that attached to Hendricks’ earlier crimes, and in a way that significantly “disadvantage^] the offender,” Weaver v. Graham, 450 U. S. 24, 29 (1981). See Brief for Respondent State of Kansas 37-39.

To find a violation of that Clause here, however, is not to hold that the Clause prevents Kansas, or other States, from enacting dangerous sexual offender statutes. A statute that operates prospectively, for example, does not offend the Ex Post Facto Clause. Weaver, 450 U. S., at 29. Neither does it offend the Ex Post Facto Clause for a State to sentence offenders to the fully authorized sentence, to seek consecutive, rather than concurrent, sentences, or to invoke recidivism statutes to lengthen imprisonment. Moreover, a statute that operates retroactively, like Kansas’ statute, nonetheless does not offend the Clause if the confinement that it imposes is not punishment — if, that is to say, the legislature does not simply add a later criminal punishment to an earlier one. Ibid.

*396The statutory provisions before us do amount to punishment primarily because, as I have said, the legislature did not tailor the statute to fit the nonpunitive civil aim of treatment, which it concedes exists in Hendricks’ case. The Clause in these circumstances does not stand as an obstacle to achieving important protections for the public’s safety; rather it provides an assurance that, where so significant a restriction of an individual’s basic freedoms is at issue, a State cannot cut corners. Rather, the legislature must hew to the Constitution’s liberty-protecting line. See The Federalist No. 78, p. 466 (C. Rossiter ed. 1961) (A. Hamilton).

I therefore would affirm the judgment below.

*397APPENDIX TO OPINION OF BREYER, J.

Selected Sexual Offense Commitment Statutes

(Kansas is the only State that answers “yes” to all three categories)

*398Selected Sexual Offense Commitment Statutes — Continued