McKUNE, WARDEN, et al.
v.
LILE

*25 *26 *27 *28 Kennedy, J., announced the judgment of the Court and delivered an opinion, in which Rehnquist, C. J., and Scalia and Thomas, JJ., joined. O'Connor, J., filed an opinion concurring in the judgment, post, p. 48. Stevens, J., filed a dissenting opinion, in which Souter, Ginsburg, and Breyer, JJ., joined, post, p. 54.

Stephen R. McAllister, State Solicitor of Kansas, argued the cause for petitioners. With him on the briefs were Carla J. Stovall, Attorney General, Jared S. Maag, and Timothy G. Madden.

Gregory G. Garre argued the cause for the United States as amicus curiae urging reversal. With him on the brief were Solicitor General Olson, Acting Assistant Attorney General Schiffer, Deputy Solicitor General Clement, and Vicki Marani.

Matthew J. Wiltanger argued the cause for respondent. With him on the brief was Paul W. Rebein.^[*]

*29 Justice Kennedy announced the judgment of the Court and delivered an opinion, in which The Chief Justice, Justice Scalia, and Justice Thomas join.

Respondent Robert G. Lile is a convicted sex offender in the custody of the Kansas Department of Corrections (Department). A few years before respondent was scheduled to reenter society, Department officials recommended that he enter a prison treatment program so that he would not rape again upon release. While there appears to be some difference of opinion among experts in the field, Kansas officials and officials who administer the United States prison system have made the determination that it is of considerable importance for the program participant to admit having committed the crime for which he is being treated and other past offenses. The first and in many ways most crucial step in the Kansas rehabilitation program thus requires the participant to confront his past crimes so that he can begin to understand his own motivations and weaknesses. As this initial step can be a most difficult one, Kansas offers sex offenders incentives to participate in the program.

Respondent contends this incentive system violates his Fifth Amendment privilege against self-incrimination. Kansas' rehabilitation program, however, serves a vital penological purpose, and offering inmates minimal incentives to participate does not amount to compelled self-incrimination prohibited by the Fifth Amendment.

I

In 1982, respondent lured a high school student into his car as she was returning home from school. At gunpoint, respondent forced the victim to perform oral sodomy on him *30 and then drove to a field where he raped her. After the sexual assault, the victim went to her school, where, crying and upset, she reported the crime. The police arrested respondent and recovered on his person the weapon he used to facilitate the crime. State v. Lile, 237 Kan. 210, 211-212, 699 P.2d 456, 457-458 (1985). Although respondent maintained that the sexual intercourse was consensual, a jury convicted him of rape, aggravated sodomy, and aggravated kidnaping. Both the Kansas Supreme Court and a Federal District Court concluded that the evidence was sufficient to sustain respondent's conviction on all charges. See id., at 211, 699 P.2d, at 458; 45 F. Supp. 2d 1157, 1161 (Kan. 1999).

In 1994, a few years before respondent was scheduled to be released, prison officials ordered him to participate in a Sexual Abuse Treatment Program (SATP). As part of the program, participating inmates are required to complete and sign an "Admission of Responsibility" form, in which they discuss and accept responsibility for the crime for which they have been sentenced. Participating inmates also are required to complete a sexual history form, which details all prior sexual activities, regardless of whether such activities constitute uncharged criminal offenses. A polygraph examination is used to verify the accuracy and completeness of the offender's sexual history.

While information obtained from participants advances the SATP's rehabilitative goals, the information is not privileged. Kansas leaves open the possibility that new evidence might be used against sex offenders in future criminal proceedings. In addition, Kansas law requires the SATP staff to report any uncharged sexual offenses involving minors to law enforcement authorities. Although there is no evidence that incriminating information has ever been disclosed under the SATP, the release of information is a possibility.

Department officials informed respondent that if he refused to participate in the SATP, his privilege status would be reduced from Level III to Level I. As part of this reduction, *31 respondent's visitation rights, earnings, work opportunities, ability to send money to family, canteen expenditures, access to a personal television, and other privileges automatically would be curtailed. In addition, respondent would be transferred to a maximum-security unit, where his movement would be more limited, he would be moved from a twoperson to a four-person cell, and he would be in a potentially more dangerous environment.

Respondent refused to participate in the SATP on the ground that the required disclosures of his criminal history would violate his Fifth Amendment privilege against selfincrimination. He brought this action under Rev. Stat. § 1979, 42 U.S. C. § 1983, against the warden and the secretary of the Department, seeking an injunction to prevent them from withdrawing his prison privileges and transferring him to a different housing unit.

After the parties completed discovery, the United States District Court for the District of Kansas entered summary judgment in respondent's favor. 24 F. Supp. 2d 1152 (1998). The District Court noted that because respondent had testified at trial that his sexual intercourse with the victim was consensual, an acknowledgment of responsibility for the rape on the "Admission of Guilt" form would subject respondent to a possible charge of perjury. Id., at 1157. After reviewing the specific loss of privileges and change in conditions of confinement that respondent would face for refusing to incriminate himself, the District Court concluded that these consequences constituted coercion in violation of the Fifth Amendment.

The Court of Appeals for the Tenth Circuit affirmed. 224 F.3d 1175 (2000). It held that the compulsion element of a Fifth Amendment claim can be established by penalties that do not constitute deprivations of protected liberty interests under the Due Process Clause. Id., at 1183. It held that the reduction in prison privileges and housing accommodations was a penalty, both because of its substantial impact *32 on the inmate and because that impact was identical to the punishment imposed by the Department for serious disciplinary infractions. In the Court of Appeals' view, the fact that the sanction was automatic, rather than conditional, supported the conclusion that it constituted compulsion. Moreover, because all SATP files are subject to disclosure by subpoena, and an admission of culpability regarding the crime of conviction would create a risk of a perjury prosecution, the court concluded that the information disclosed by respondent was sufficiently incriminating. Id., at 1180. The Court of Appeals recognized that the Kansas policy served the State's important interests in rehabilitating sex offenders and promoting public safety. It concluded, however, that those interests could be served without violating the Constitution, either by treating the admissions of the inmates as privileged communications or by granting inmates use immunity. Id., at 1192.

We granted the warden's petition for certiorari because the Court of Appeals has held that an important Kansas prison regulation violates the Federal Constitution. 532 U.S. 1018 (2001).

II

Sex offenders are a serious threat in this Nation. In 1995, an estimated 355,000 rapes and sexual assaults occurred nationwide. U. S. Dept. of Justice, Bureau of Justice Statistics, Sex Offenses and Offenders 1 (1997) (hereinafter Sex Offenses); U. S. Dept. of Justice, Federal Bureau of Investigation, Crime in the United States, 1999, Uniform Crime Reports 24 (2000). Between 1980 and 1994, the population of imprisoned sex offenders increased at a faster rate than for any other category of violent crime. See Sex Offenses 18. As in the present case, the victims of sexual assault are most often juveniles. In 1995, for instance, a majority of reported forcible sexual offenses were committed against persons under 18 years of age. University of New Hampshire, Crimes Against Children Research Center, Fact Sheet 5; Sex Offenses 24. Nearly 4 in 10 imprisoned violent *33 sex offenders said their victims were 12 or younger. Id., at iii.

When convicted sex offenders reenter society, they are much more likely than any other type of offender to be rearrested for a new rape or sexual assault. See id., at 27; U. S. Dept. of Justice, Bureau of Justice Statistics, Recidivism of Prisoners Released in 1983, p. 6 (1997). States thus have a vital interest in rehabilitating convicted sex offenders.

Therapists and correctional officers widely agree that clinical rehabilitative programs can enable sex offenders to manage their impulses and in this way reduce recidivism. See U. S. Dept. of Justice, Nat. Institute of Corrections, A Practitioner's Guide to Treating the Incarcerated Male Sex Offender xiii(1988) ("[T]he rate of recidivism of treated sex offenders is fairly consistently estimated to be around 15%," whereas the rate of recidivism of untreated offenders has been estimated to be as high as 80%. "Even if both of these figures are exaggerated, there would still be a significant difference between treated and untreated individuals"). An important component of those rehabilitation programs requires participants to confront their past and accept responsibility for their misconduct. Id., at 73. "Denial is generally regarded as a main impediment to successful therapy," and "[t]herapists depend on offenders' truthful descriptions of events leading to past offences in order to determine which behaviours need to be targeted in therapy." H. Barbaree, Denial and Minimization Among Sex Offenders: Assessment and Treatment Outcome, 3 Forum on Corrections Research, No. 4, p. 30 (1991). Research indicates that offenders who deny all allegations of sexual abuse are three times more likely to fail in treatment than those who admit even partial complicity. See B. Maletzky & K. McGovern, Treating the Sexual Offender 253-255 (1991).

The critical first step in the Kansas SATP, therefore, is acceptance of responsibility for past offenses. This gives inmates a basis to understand why they are being punished *34 and to identify the traits that cause such a frightening and high risk of recidivism. As part of this first step, Kansas requires each SATP participant to complete an "Admission of Responsibility" form, to fill out a sexual history form discussing their offending behavior, and to discuss their past behavior in individual and group counseling sessions.

The District Court found that the Kansas SATP is a valid "clinical rehabilitative program," supported by a "legitimate penological objective" in rehabilitation. 24 F. Supp. 2d, at 1163. The SATP lasts for 18 months and involves substantial daily counseling. It helps inmates address sexual addiction; understand the thoughts, feelings, and behavior dynamics that precede their offenses; and develop relapse prevention skills. Although inmates are assured of a significant level of confidentiality, Kansas does not offer legal immunity from prosecution based on any statements made in the course of the SATP. According to Kansas, however, no inmate has ever been charged or prosecuted for any offense based on information disclosed during treatment. Brief for Petitioners 4-5. There is no contention, then, that the program is a mere subterfuge for the conduct of a criminal investigation.

As the parties explain, Kansas' decision not to offer immunity to every SATP participant serves two legitimate state interests. First, the professionals who design and conduct the program have concluded that for SATP participants to accept full responsibility for their past actions, they must accept the proposition that those actions carry consequences. Tr. of Oral Arg. 11. Although no program participant has ever been prosecuted or penalized based on information revealed during the SATP, the potential for additional punishment reinforces the gravity of the participants' offenses and thereby aids in their rehabilitation. If inmates know society will not punish them for their past offenses, they may be left with the false impression that society does not consider those crimes to be serious ones. The practical effect of guaranteed *35 immunity for SATP participants would be to absolve many sex offenders of any and all cost for their earlier crimes. This is the precise opposite of the rehabilitative objective.

Second, while Kansas as a rule does not prosecute inmates based upon information revealed in the course of the program, the State confirms its valid interest in deterrence by keeping open the option to prosecute a particularly dangerous sex offender. Brief for 18 States as Amici Curiae 11. Kansas is not alone in declining to offer blanket use immunity as a condition of participation in a treatment program. The Federal Bureau of Prisons and other States conduct similar sex offender programs and do not offer immunity to the participants. See, e. g., Ainsworth v. Risley, 244 F.3d 209, 214 (CA1 2001) (describing New Hampshire's program).

The mere fact that Kansas declines to grant inmates use immunity does not render the SATP invalid. Asking at the outset whether prison administrators can or should offer immunity skips the constitutional inquiry altogether. If the State of Kansas offered immunity, the self-incrimination privilege would not be implicated. See, e. g., Kastigar v. United States, 406 U.S. 441, 453 (1972); Brown v. Walker, 161 U.S. 591, 610 (1896). The State, however, does not offer immunity. So the central question becomes whether the State's program, and the consequences for nonparticipation in it, combine to create a compulsion that encumbers the constitutional right. If there is compulsion, the State cannot continue the program in its present form; and the alternatives, as will be discussed, defeat the program's objectives.

The SATP does not compel prisoners to incriminate themselves in violation of the Constitution. The Fifth Amendment Self-Incrimination Clause, which applies to the States via the Fourteenth Amendment, Malloy v. Hogan, 378 U.S. 1 (1964), provides that no person "shall be compelled in any criminal case to be a witness against himself." The "Amendment speaks of compulsion," United States v. Monia, *36 317 U.S. 424, 427 (1943), and the Court has insisted that the "constitutional guarantee is only that the witness not be compelled to give self-incriminating testimony." United States v. Washington, 431 U.S. 181, 188 (1977). The consequences in question here—a transfer to another prison where television sets are not placed in each inmate's cell, where exercise facilities are not readily available, and where work and wage opportunities are more limited—are not ones that compel a prisoner to speak about his past crimes despite a desire to remain silent. The fact that these consequences are imposed on prisoners, rather than ordinary citizens, moreover, is important in weighing respondent's constitutional claim.

The privilege against self-incrimination does not terminate at the jail house door, but the fact of a valid conviction and the ensuing restrictions on liberty are essential to the Fifth Amendment analysis. Sandin v. Conner, 515 U.S. 472, 485 (1995) ("[L]awful incarceration brings about the necessary withdrawal or limitation of many privileges and rights, a retraction justified by the considerations underlying our penal system" (citation and internal quotation marks omitted)). A broad range of choices that might infringe constitutional rights in a free society fall within the expected conditions of confinement of those who have suffered a lawful conviction.

The Court has instructed that rehabilitation is a legitimate penological interest that must be weighed against the exercise of an inmate's liberty. See, e. g., O'Lone v. Estate of Shabazz, 482 U.S. 342, 348, 351 (1987). Since "most offenders will eventually return to society, [a] paramount objective of the corrections system is the rehabilitation of those committed to its custody." Pell v. Procunier, 417 U.S. 817, 823 (1974). Acceptance of responsibility in turn demonstrates that an offender "is ready and willing to admit his crime and to enter the correctional system in a frame of mind that affords hope for success in rehabilitation over a shorter period *37 of time than might otherwise be necessary." Brady v. United States, 397 U.S. 742, 753 (1970).

The limitation on prisoners' privileges and rights also follows from the need to grant necessary authority and capacity to federal and state officials to administer the prisons. See, e. g., Turner v. Safley, 482 U.S. 78 (1987). "Running a prison is an inordinately difficult undertaking that requires expertise, planning, and the commitment of resources, all of which are peculiarly within the province of the legislative and executive branches of government." Id., at 84-85. To respect these imperatives, courts must exercise restraint in supervising the minutiae of prison life. Ibid. Where, as here, a state penal system is involved, federal courts have "additional reason to accord deference to the appropriate prison authorities." Ibid.

For these reasons, the Court in Sandin held that challenged prison conditions cannot give rise to a due process violation unless those conditions constitute "atypical and significant hardship[s] on [inmates] in relation to the ordinary incidents of prison life." See 515 U.S., at 484. The determination under Sandin whether a prisoner's liberty interest has been curtailed may not provide a precise parallel for determining whether there is compelled self-incrimination, but it does provide useful instruction for answering the latter inquiry. Sandin and its counterparts underscore the axiom that a convicted felon's life in prison differs from that of an ordinary citizen. In the context of a legitimate rehabilitation program for prisoners, those same considerations are relevant to our analysis. The compulsion inquiry must consider the significant restraints already inherent in prison life and the State's own vital interests in rehabilitation goals and procedures within the prison system. A prison clinical rehabilitation program, which is acknowledged to bear a rational relation to a legitimate penological objective, does not violate the privilege against self-incrimination if the adverse *38 consequences an inmate faces for not participating are related to the program objectives and do not constitute atypical and significant hardships in relation to the ordinary incidents of prison life.

Along these lines, this Court has recognized that lawful conviction and incarceration necessarily place limitations on the exercise of a defendant's privilege against selfincrimination. See, e. g., Baxter v. Palmigiano, 425 U.S. 308 (1976). Baxter declined to extend to prison disciplinary proceedings the rule of Griffin v. California, 380 U.S. 609 (1965), that the prosecution may not comment on a defendant's silence at trial. 425 U.S., at 319-320. As the Court explained, "[d]isciplinary proceedings in state prisons . . . involve the correctional process and important state interests other than conviction for crime." Id., at 319. The inmate in Baxter no doubt felt compelled to speak in one sense of the word. The Court, considering the level of compulsion in light of the prison setting and the State's interests in rehabilitation and orderly administration, nevertheless rejected the inmate's self-incrimination claim.

In the present case, respondent's decision not to participate in the Kansas SATP did not extend his term of incarceration. Nor did his decision affect his eligibility for goodtime credits or parole. 224 F.3d, at 1182. Respondent instead complains that if he remains silent about his past crimes, he will be transferred from the medium-security unit—where the program is conducted—to a less desirable maximum-security unit.

No one contends, however, that the transfer is intended to punish prisoners for exercising their Fifth Amendment rights. Rather, the limitation on these rights is incidental to Kansas' legitimate penological reason for the transfer: Due to limited space, inmates who do not participate in their respective programs will be moved out of the facility where the programs are held to make room for other inmates. As the Secretary of Corrections has explained, "it makes no *39 sense to have someone who's not participating in a program taking up a bed in a setting where someone else who may be willing to participate in a program could occupy that bed and participate in a program." App. 99.

It is well settled that the decision where to house inmates is at the core of prison administrators' expertise. See Meachum v. Fano, 427 U.S. 215, 225 (1976). For this reason the Court has not required administrators to conduct a hearing before transferring a prisoner to a bed in a different prison, even if "life in one prison is much more disagreeable than in another." Ibid. The Court has considered the proposition that a prisoner in a more comfortable facility might begin to feel entitled to remain there throughout his term of incarceration. The Court has concluded, nevertheless, that this expectation "is too ephemeral and insubstantial to trigger procedural due process protections as long as prison officials have discretion to transfer him for whatever reason or for no reason at all." Id., at 228. This logic has equal force in analyzing respodent's self-incrimination claim.

Respondent also complains that he will be demoted from Level III to Level I status as a result of his decision not to participate. This demotion means the loss of his personal television; less access to prison organizations and the gym area; a reduction in certain pay opportunities and canteen privileges; and restricted visitation rights. App. 27-28. An essential tool of prison administration, however, is the authority to offer inmates various incentives to behave. The Constitution accords prison officials wide latitude to bestow or revoke these perquisites as they see fit. Accordingly, Hewitt v. Helms, 459 U.S. 460, 467, n. 4 (1983), held that an inmate's transfer to another facility did not in itself implicate a liberty interest, even though that transfer resulted in the loss of "access to vocational, educational, recreational, and rehabilitative programs." Respondent concedes that no liberty interest is implicated in this case. Tr. of Oral Arg. 45. To be sure, cases like Meachum and *40 Hewitt involved the Due Process Clause rather than the privilege against compelled self-incrimination. Those cases nevertheless underscore the axiom that, by virtue of their convictions, inmates must expect significant restrictions, inherent in prison life, on rights and privileges free citizens take for granted.

Respondent fails to cite a single case from this Court holding that the denial of discrete prison privileges for refusal to participate in a rehabilitation program amounts to unconstitutional compulsion. Instead, relying on the so-called penalty cases, respondent treats the fact of his incarceration as if it were irrelevant. See, e. g., Garrity v. New Jersey, 385 U.S. 493 (1967); Spevack v. Klein, 385 U.S. 511 (1967). Those cases, however, involved free citizens given the choice between invoking the Fifth Amendment privilege and sustaining their economic livelihood. See, e. g., id., at 516 ("[T]hreat of disbarment and the loss of professional standing, professional reputation, and of livelihood are powerful forms of compulsion"). Those principles are not easily extended to the prison context, where inmates surrender upon incarceration their rights to pursue a livelihood and to contract freely with the State, as well as many other basic freedoms. The persons who asserted rights in Garrity and Spevack had not been convicted of a crime. It would come as a surprise if Spevack stands for the proposition that when a lawyer has been disbarred by reason of a final criminal conviction, the court or agency considering reinstatement of the right to practice law could not consider that the disbarred attorney has admitted his guilt and expressed contrition. Indeed, this consideration is often given dispositive weight by this Court itself on routine motions for reinstatement. The current case is more complex, of course, in that respondent is also required to discuss other criminal acts for which he might still be liable for prosecution. On this point, however, there is still a critical distinction between the instant case and Garrity or Spevack. Unlike those cases, *41 respondent here is asked to discuss other past crimes as part of a legitimate rehabilitative program conducted within prison walls.

To reject out of hand these considerations would be to ignore the State's interests in offering rehabilitation programs and providing for the efficient administration of its prisons. There is no indication that the SATP is an elaborate attempt to avoid the protections offered by the privilege against compelled self-incrimination. Rather, the program serves an important social purpose. It would be bitter medicine to treat as irrelevant the State's legitimate interests and to invalidate the SATP on the ground that it incidentally burdens an inmate's right to remain silent.

Determining what constitutes unconstitutional compulsion involves a question of judgment: Courts must decide whether the consequences of an inmate's choice to remain silent are closer to the physical torture against which the Constitution clearly protects or the de minimis harms against which it does not. The Sandin framework provides a reasonable means of assessing whether the response of prison administrators to correctional and rehabilitative necessities are so out of the ordinary that one could sensibly say they rise to the level of unconstitutional compulsion.

Prison context or not, respondent's choice is marked less by compulsion than by choices the Court has held give no rise to a self-incrimination claim. The "criminal process, like the rest of the legal system, is replete with situations requiring the making of difficult judgments as to which course to follow. Although a defendant may have a right, even of constitutional dimensions, to follow whichever course he chooses, the Constitution does not by that token always forbid requiring him to choose." McGautha v. California, 402 U.S. 183, 213 (1971) (citation and internal quotation marks omitted). It is well settled that the government need not make the exercise of the Fifth Amendment privilege cost free. See, e. g., Jenkins v. Anderson, 447 U.S. 231, 238 *42 (1980) (a criminal defendant's exercise of his Fifth Amendment privilege prior to arrest may be used to impeach his credibility at trial); Williams v. Florida, 399 U.S. 78, 84-85 (1970) (a criminal defendant may be compelled to disclose the substance of an alibi defense prior to trial or be barred from asserting it).

The cost to respondent of exercising his Fifth Amendment privilege—denial of certain perquisites that make his life in prison more tolerable—is much less than that borne by the defendant in McGautha. There, the Court upheld a procedure that allowed statements, which were made by a criminal defendant to mitigate his responsibility and avoid the death penalty, to be used against him as evidence of his guilt. 402 U.S., at 217. The Court likewise has held that plea bargaining does not violate the Fifth Amendment, even though criminal defendants may feel considerable pressure to admit guilt in order to obtain more lenient treatment. See, e. g., Bordenkircher v. Hayes, 434 U.S. 357 (1978); Brady, 397 U. S., at 751.

Nor does reducing an inmate's prison wage and taking away personal television and gym access pose the same hard choice faced by the defendants in Baxter v. Palmigiano, 425 U.S. 308 (1976), Ohio Adult Parole Authority v. Woodard, 523 U.S. 272 (1998), and Minnesota v. Murphy, 465 U.S. 420 (1984). In Baxter, a state prisoner objected to the fact that his silence at a prison disciplinary hearing would be held against him. The Court acknowledged that Griffin v. California, 380 U.S. 609 (1965), held that the Fifth Amendment prohibits courts from instructing a criminal jury that it may draw an inference of guilt from a defendant's failure to testify. The Court nevertheless refused to extend the Griffin rule to the context of state prison disciplinary hearings because those proceedings "involve the correctional process and important state interests other than conviction for crime." 425 U.S., at 319. Whereas the inmate in the present case faces the loss of certain privileges, the prisoner in *43 Baxter faced 30 days in punitive segregation as well as the subsequent downgrade of his prison classification status. Id., at 313.

In Murphy, the defendant feared the possibility of additional jail time as a result of his decision to remain silent. The defendant's probation officer knew the defendant had committed a rape and murder unrelated to his probation. One of the terms of the defendant's probation required him to be truthful with the probation officer in all matters. Seizing upon this, the officer interviewed the defendant about the rape and murder, and the defendant admitted his guilt. The Court found no Fifth Amendment violation, despite the defendant's fear of being returned to prison for 16 months if he remained silent. 465 U.S., at 422, 438.

In Woodard, the plaintiff faced not loss of a personal television and gym access, but loss of life. In a unanimous opinion just four Terms ago, this Court held that a death row inmate could be made to choose between incriminating himself at his clemency interview and having adverse inferences drawn from his silence. The Court reasoned that it "is difficult to see how a voluntary interview could `compel' respondent to speak. He merely faces a choice quite similar to the sorts of choices that a criminal defendant must make in the course of criminal proceedings, none of which has ever been held to violate the Fifth Amendment." 523 U.S., at 286. As here, the inmate in Woodard claimed to face a Hobson's choice: He would damage his case for clemency no matter whether he spoke and incriminated himself, or remained silent and the clemency board construed that silence against him. Unlike here, the Court nevertheless concluded that the pressure the inmate felt to speak to improve his chances of clemency did not constitute unconstitutional compulsion. Id., at 287-288.

Woodard, Murphy, and Baxter illustrate that the consequences respondent faced here did not amount to unconstitutional compulsion. Respondent and the dissent attempt to distinguish Baxter, Murphy, and Woodard on the dual *44 grounds that (1) the penalty here followed automatically from respondent's decision to remain silent, and (2) respondent's participation in the SATP was involuntary. Neither distinction would justify departing from this Court's precedents, and the second is question begging in any event.

It is proper to consider the nexus between remaining silent and the consequences that follow. Plea bargains are not deemed to be compelled in part because a defendant who pleads not guilty still must be convicted. Cf. Brady, supra, at 751-752. States may award good-time credits and early parole for inmates who accept responsibility because silence in these circumstances does not automatically mean the parole board, which considers other factors as well, will deny them parole. See Baxter, supra, at 317-318. While the automatic nature of the consequence may be a necessary condition to finding unconstitutional compulsion, however, that is not a sufficient reason alone to ignore Woodard, Murphy, and Baxter. Even if a consequence follows directly from a person's silence, one cannot answer the question whether the person has been compelled to incriminate himself without first considering the severity of the consequences.

Nor can Woodard be distinguished on the alternative ground that respondent's choice to participate in the SATP was involuntary, whereas the death row inmate in Woodard chose to participate in clemency proceedings. This distinction assumes the answer to the compulsion inquiry. If respondent was not compelled to participate in the SATP, his participation was voluntary in the only sense necessary for our present inquiry. Kansas asks sex offenders to participate in SATP because, in light of the high rate of recidivism, it wants all, not just the few who volunteer, to receive treatment. Whether the inmates are being asked or ordered to participate depends entirely on the consequences of their decision not to do so. The parties in Woodard, Murphy, and Baxter all were faced with ramifications far worse than respondent faces here, and in each of those cases the Court *45 determined that their hard choice between silence and the consequences was not compelled. It is beyond doubt, of course, that respondent would prefer not to choose between losing prison privileges and accepting responsibility for his past crimes. It is a choice, nonetheless, that does not amount to compulsion, and therefore one Kansas may require respondent to make.

The Federal Government has filed an amicus brief describing its sex offender treatment program. Were respondent's position to prevail, the constitutionality of the federal program would be cast into serious doubt. The fact that the offender in the federal program can choose to participate without being given a new prisoner classification is not determinative. For, as the Government explains, its program is conducted at a single, 112-bed facility that is more desirable than other federal prisons. Tr. of Oral Arg. 22. Inmates choose at the outset whether to enter the federal program. Once accepted, however, inmates must continue to discuss and accept responsibility for their crimes if they wish to maintain the status quo and remain in their more comfortable accommodations. Otherwise they will be expelled from the program and sent to a less desirable facility. Id., at 27. Thus the federal program is different from Kansas' SATP only in that it does not require inmates to sacrifice privileges besides housing as a consequence of nonparticipation. The federal program is comparable to the Kansas program because it does not offer participants use immunity and because it conditions a desirable housing assignment on inmates' willingness to accept responsibility for past behavior. Respondent's theory cannot be confined in any meaningful way, and state and federal courts applying that view would have no principled means to determine whether these similarities are sufficient to render the federal program unconstitutional.

Respondent is mistaken as well to concentrate on the socalled reward/penalty distinction and the illusory baseline *46 against which a change in prison conditions must be measured. The answer to the question whether the government is extending a benefit or taking away a privilege rests entirely in the eye of the be holder. For this reason, emphasis of any baseline, while superficially appealing, would be an in artful addition to an already confused area of jurisprudence. The prison warden in this case stated that it is largely a matter of chance where in a prison an inmate is assigned. App. 59-63. Even if Inmates A and B are serving the same sentence for the same crime, Inmate A could end up in a medium-security unit and Inmate B in a maximum-security unit based solely on administrative factors beyond their control. Under respondent's view, however, the Constitution allows the State to offer Inmate B the opportunity to live in the medium-security unit conditioned on his participation in the SATP, but does not allow the State to offer Inmate A the opportunity to live in that same medium-security unit subject to the same conditions. The consequences for Inmates A and B are identical: They may participate and live in medium security or refuse and live in maximum security. Respondent, however, would have us say the Constitution puts Inmate A in a superior position to Inmate B solely by the accident of the initial assignment to a medium-security unit.

This reasoning is unsatisfactory. The Court has noted before that "[w]e doubt that a principled distinction may be drawn between `enhancing' the punishment imposed upon the petitioner and denying him the `leniency' he claims would be appropriate if he had cooperated." Roberts v. United States, 445 U.S. 552, 557, n. 4 (1980). Respondent's reasoning would provide States with perverse incentives to assign all inmates convicted of sex offenses to maximum security prisons until near the time of release, when the rehabilitation program starts. The rule would work to the detriment of the entire class of sex offenders who might not otherwise be placed in maximum-security facilities. And prison administrators *47 would be forced, before making routine prison housing decisions, to identify each inmate's so-called baseline and determine whether an adverse effect, however marginal, will result from the administrative decision. The easy alternatives that respondent predicts for prison administrators would turn out to be not so trouble free.

Respondent's analysis also would call into question the constitutionality of an accepted feature of federal criminal law: the downward adjustment for acceptance of criminal responsibility provided in § 3E1.1 of the United States Sentencing Commission, Guidelines Manual (Nov. 2002). If the Constitution does not permit the government to condition the use of a personal television on the acceptance of responsibility for past crimes, it is unclear how it could permit the government to reduce the length of a prisoner's term of incarceration based upon the same factor. By rejecting respondent's theory, we do not, in this case, call these policies into question.

* * *

Acceptance of responsibility is the beginning of rehabilitation. And a recognition that there are rewards for those who attempt to reform is a vital and necessary step toward completion. The Court of Appeals' ruling would defeat these objectives. If the State sought to comply with the ruling by allowing respondent to enter the program while still insisting on his innocence, there would be little incentive for other SATP participants to confess and accept counseling; indeed, there is support for Kansas' view that the dynamics of the group therapy would be impaired. If the State had to offer immunity, the practical effect would be that serial offenders who are incarcerated for but one violation would be given a windfall for past bad conduct, a result potentially destructive of any public or state support for the program and quite at odds with the dominant goal of acceptance of responsibility. If the State found it was forced to graduate prisoners from its rehabilitation program without knowing *48 what other offenses they may have committed, the integrity of its program would be very much in doubt. If the State found it had to comply by allowing respondent the same perquisites as those who accept counseling, the result would be a dramatic illustration that obduracy has the same rewards as acceptance, and so the program itself would become selfdefeating, even hypocritical, in the eyes of those whom it seeks to help. The Fifth Amendment does not require the State to suffer these programmatic disruptions when it seeks to rehabilitate those who are incarcerated for valid, final convictions.

The Kansas SATP represents a sensible approach to reducing the serious danger that repeat sex offenders pose to many innocent persons, most often children. The State's interest in rehabilitation is undeniable. There is, furthermore, no indication that the SATP is merely an elaborate ruse to skirt the protections of the privilege against compelled self-incrimination. Rather, the program allows prison administrators to provide to those who need treatment the incentive to seek it.

The judgment of the Court of Appeals is reversed, and the case is remanded for further proceedings.

It is so ordered.

Justice O'Connor, concurring in the judgment.

The Court today is divided on the question of what standard to apply when evaluating compulsion for the purposes of the Fifth Amendment privilege against self-incrimination in a prison setting. I write separately because, although I agree with Justice Stevens that the Fifth Amendment compulsion standard is broader than the "atypical and significant hardship" standard we have adopted for evaluating due process claims in prisons, see post, at 58-60 (dissenting opinion) (citing Meachum v. Fano, 427 U.S. 215 (1976)), I do not believe that the alterations in respondent's prison conditions as a result of his failure to participate in the Sexual *49 Abuse Treatment Program (SATP) were so great as to constitute compulsion for the purposes of the Fifth Amendment privilege against self-incrimination. I therefore agree with the plurality that the decision below should be reversed.

The text of the Fifth Amendment does not prohibit all penalties levied in response to a person's refusal to incriminate himself or herself—it prohibits only the compulsion of such testimony. Not all pressure necessarily "compel[s]" incriminating statements.

For instance, in Miranda v. Arizona, 384 U.S. 436, 455 (1966), we found that an environment of police custodial interrogation was coercive enough to require prophylactic warnings only after observing that such an environment exerts a "heavy toll on individual liberty." But we have not required Miranda warnings during noncustodial police questioning. See, e. g., Beckwith v. United States, 425 U.S. 341 (1976). In restricting Miranda `s applicability, we have not denied that noncustodial questioning imposes some sort of pressure on suspects to confess to their crimes. See Oregon v. Mathiason, 429 U.S. 492, 495 (1977) (per curiam) ("Any interview of one suspected of a crime by a police officer will have coercive aspects to it . . ."); Berkemer v. McCarty, 468 U.S. 420, 440 (1984) (describing the "comparatively nonthreatening character of [noncustodial] detentions" (emphasis added)). Rather, as suggested by the text of the Fifth Amendment, we have asked whether the pressure imposed in such situations rises to a level where it is likely to "compe[l]" a person "to be a witness against himself."

The same analysis applies to penalties imposed upon a person as a result of the failure to incriminate himself—some penalties are so great as to "compe[l]" such testimony, while others do not rise to that level. Our precedents establish that certain types of penalties are capable of coercing incriminating testimony: termination of employment, Uniformed Sanitation Men Assn., Inc. v. Commissioner of Sanitation of City of New York, 392 U.S. 280 (1968), the loss of a professional *50 license, Spevack v. Klein, 385 U.S. 511 (1967), ineligibility to receive government contracts, Lefkowitz v. Turley, 414 U.S. 70 (1973), and the loss of the right to participate in political associations and to hold public office, Lefkowitz v. Cunningham, 431 U.S. 801 (1977). All of these penalties, however, are far more significant than those facing respondent here.

The first three of these so-called "penalty cases" involved the potential loss of one's livelihood, either through the loss of employment, loss of a professional license essential to employment, or loss of business through government contracts. In Lefkowitz, we held that the loss of government contracts was constitutionally equivalent to the loss of a profession because "[a government contractor] lives off his contracting fees just as surely as a state employee lives off his salary." 414 U.S., at 83; contra, post, at 68, n. 11. To support oneself in one's chosen profession is one of the most important abilities a person can have. A choice between incriminating oneself and being deprived of one's livelihood is the very sort of choice that is likely to compel someone to be a witness against himself. The choice presented in the last case, Cunningham, implicated not only political influence and prestige, but also the First Amendment right to run for office and to participate in political associations. 431 U.S., at 807-808. In holding that the penalties in that case constituted compulsion for Fifth Amendment purposes, we properly referred to those consequences as "grave." Id., at 807.

I do not believe the consequences facing respondent in this case are serious enough to compel him to be a witness against himself. These consequences involve a reduction in incentive level, and a corresponding transfer from a medium-security to a maximum-security part of the prison. In practical terms, these changes involve restrictions on the personal property respondent can keep in his cell, a reduction in his visitation privileges, a reduction in the amount of money he can spend in the canteen, and a reduction in the *51 wage he can earn through prison employment. See ante, at 30-31. These changes in living conditions seem to me minor. Because the prison is responsible for caring for respondent's basic needs, his ability to support himself is not implicated by the reduction in wages he would suffer as a result. While his visitation is reduced as a result of his failure to incriminate himself, he still retains the ability to see his attorney, his family, and members of the clergy. App. 27. The limitation on the possession of personal items, as well as the amount that respondent is allowed to spend at the canteen, may make his prison experience more unpleasant, but seems very unlikely to actually compel him to incriminate himself.

Justice Stevens also suggests that the move to the maximum-security area of the prison would itself be coercive. See post, at 63-64. Although the District Court found that moving respondent to a maximum-security section of the prison would put him "in a more dangerous environment occupied by more serious offenders," 24 F. Supp. 2d 1152, 1155 (Kan. 1998), there was no finding about how great a danger such a placement posed. Because it is respondent's burden to prove compulsion, we may assume that the prison is capable of controlling its inmates so that respondent's personal safety is not jeopardized by being placed in the maximum-security area of the prison, at least in the absence of proof to the contrary.

Justice Stevens argues that the fact that the penalties facing respondent for refusal to incriminate himself are the same as those imposed for prison disciplinary violations also indicates that they are coercive. See post, at 62-63. I do not agree. Insofar as Justice Stevens' claim is that these sanctions carry a stigma that might compel respondent to incriminate himself, it is incorrect. Because the same sanctions are also imposed on all prisoners who refuse to participate in any recommended program, App. 19-20, any stigma attached to the reduction would be minimal. Insofar as *52 Justice Stevens' claim is that these sanctions are designed to compel behavior because they are used as disciplinary tools, it is also flawed. There is a difference between the sorts of penalties that would give a prisoner a reason not to violate prison disciplinary rules and what would compel him to expose himself to criminal liability. Therefore, on this record, I cannot conclude that respondent has shown that his decision to incriminate himself would be compelled by the imposition of these penalties.

Although I do not think the penalties respondent faced were sufficiently serious to compel his testimony, I do not agree with the suggestion in the plurality opinion that these penalties could permissibly rise to the level of those in cases like McGautha v. California, 402 U.S. 183 (1971) (holding that statements made in the mitigation phase of a capital sentencing hearing may be used as evidence of guilt), Bordenkircher v. Hayes, 434 U.S. 357 (1978) (holding that plea bargaining does not violate the Fifth Amendment privilege against self-incrimination), and Ohio Adult Parole Authority v. Woodard, 523 U.S. 272 (1998) (holding that there is no right to silence at a clemency interview). See ante, at 41-43. The penalties potentially faced in these cases—longer incarceration and execution—are far greater than those we have already held to constitute unconstitutional compulsion in the penalty cases. Indeed, the imposition of such outcomes as a penalty for refusing to incriminate oneself would surely implicate a "liberty interest."

Justice Stevens attempts to distinguish these cases because, in each, the negative outcome did not follow directly from the decision to remain silent, and because none of these cases involved a direct order to testify. See post, at 60. As the plurality's opinion makes clear, however, these two factors do not adequately explain the difference between these cases and the penalty cases, where we have found compulsion based on the imposition of penalties far less onerous. See ante, at 43-45.

*53 I believe the proper theory should recognize that it is generally acceptable to impose the risk of punishment, however great, so long as the actual imposition of such punishment is accomplished through a fair criminal process. See, e. g., McGautha v. California, supra, at 213 ("The criminal process, like the rest of the legal system, is replete with situations requiring the making of difficult judgments as to which course to follow. Although a defendant may have a right, even of constitutional dimensions, to follow whichever course he chooses, the Constitution does not by that token always forbid requiring him to choose" (citation and internal quotation marks omitted)). Forcing defendants to accept such consequences seems to me very different from imposing penalties for the refusal to incriminate oneself that go beyond the criminal process and appear, starkly, as government attempts to compel testimony; in the latter context, any penalty that is capable of compelling a person to be a witness against himself is illegitimate. But even this explanation of the privilege is incomplete, as it does not fully account for all of the Court's precedents in this area. Compare Griffin v. California, 380 U.S. 609 (1965) (holding that prosecutor may not comment on a defendant's failure to testify), with Ohio Adult Parole Authority v. Woodard, supra (holding that there is no right to silence at a clemency interview).

Complicating matters even further is the question of whether the denial of benefits and the imposition of burdens ought to be analyzed differently in this area. Compare ante, at 45-47, with post, at 64-65. This question is particularly important given the existence of United States Sentencing Commission, Guidelines Manual § 3E1.1 (Nov. 2000), which can be read to offer convicted criminals the benefit of a lower sentence in exchange for accepting responsibility for their crimes. See ante, at 47.

I find the plurality's failure to set forth a comprehensive theory of the Fifth Amendment privilege against selfincrimination troubling. But because this case indisputably *54 involves burdens rather than benefits, and because I do not believe the penalties assessed against respondent in response to his failure to incriminate himself are compulsive on any reasonable test, I need not resolve this dilemma to make my judgment in this case.

Although I do not agree that the standard for compulsion is the same as the due process standard we identified in Sandin v. Conner, 515 U.S. 472 (1995), I join in the judgment reached by the plurality's opinion.

Justice Stevens, with whom Justice Souter, Justice Ginsburg, and Justice Breyer join, dissenting.

No one could possibly disagree with the plurality's statement that "offering inmates minimal incentives to participate [in a rehabilitation program] does not amount to compelled self-incrimination prohibited by the Fifth Amendment." Ante, at 29. The question that this case presents, however, is whether the State may punish an inmate's assertion of his Fifth Amendment privilege with the same mandatory sanction that follows a disciplinary conviction for an offense such as theft, sodomy, riot, arson, or assault. Until today the Court has never characterized a threatened harm as "a minimal incentive." Nor have we ever held that a person who has made a valid assertion of the privilege may nevertheless be ordered to incriminate himself and sanctioned for disobeying such an order. This is truly a watershed case.

Based on an ad hoc appraisal of the benefits of obtaining confessions from sex offenders, balanced against the cost of honoring a bedrock constitutional right, the plurality holds that it is permissible to punish the assertion of the privilege with what it views as modest sanctions, provided that those sanctions are not given a "punitive" label. As I shall explain, the sanctions are in fact severe, but even if that were not so, the plurality's policy judgment does not justify the evisceration of a constitutional right. Despite the plurality's *55 meandering attempt to justify its unprecedented departure from a rule of law that has been settled since the days of John Marshall, I respectfully dissent.

I

The text of the Fifth Amendment provides that no person "shall be compelled in any criminal case to be a witness against himself." It is well settled that the prohibition "not only permits a person to refuse to testify against himself at a criminal trial in which he is a defendant, but also `privileges him not to answer official questions put to him in any other proceeding, civil or criminal, formal or informal, where the answers might incriminate him in future criminal proceedings.' " Minnesota v. Murphy, 465 U.S. 420, 426 (1984) (quoting Lefkowitz v. Turley, 414 U.S. 70, 77 (1973)). If a person is protected by the privilege, he may "refuse to answer unless and until he is protected at least against the use of his compelled answers and evidence derived therefrom in any subsequent criminal case in which he is a defendant." Id., at 78 (citing Kastigar v. United States, 406 U.S. 441 (1972)). Prison inmates—including sex offenders—do not forfeit the privilege at the jail house gate. Murphy, 465 U. S., at 426.

It is undisputed that respondent's statements on the admission of responsibility and sexual history forms could incriminate him in a future prosecution for perjury or any other offense to which he is forced to confess.^[1] It is also *56 clear that he invoked his Fifth Amendment right by refusing to participate in the SATP on the ground that he would be required to incriminate himself. Once he asserted that right, the State could have offered respondent immunity from the use of his statements in a subsequent prosecution. Instead, the Kansas Department of Corrections (Department) ordered respondent either to incriminate himself or to lose his medium-security status. In my opinion that order, coupled with the threatened revocation of respondent's Level III privileges, unquestionably violated his Fifth Amendment rights.

Putting to one side the plurality's evaluation of the policy judgments made by Kansas, its central submission is that the threatened withdrawal of respondent's Level III and medium-security status is not sufficiently harmful to qualify as unconstitutional compulsion. In support of this position, neither the plurality nor Justice O'Connor cites a single Fifth Amendment case in which a person invoked the privilege and was nevertheless required to answer a potentially incriminating question.^[2]

The privilege against self-incrimination may have been born of the rack and the Star Chamber, see L. Levy, Origins of the Fifth Amendment 42 (I. Dee ed. 1999); Andresen v. Maryland, 427 U.S. 463, 470 (1976), but the Framers had a *57 broader view of compulsion in mind when they drafted the Fifth Amendment.^[3] We know, for example, that the privilege was thought to protect defendants from the moral compulsion associated with any statement made under oath.^[4] In addition, the language of the Amendment, which focuses on a courtroom setting in which a defendant or a witness in a criminal trial invokes the privilege, encompasses the compulsion inherent in any judicial order overruling an assertion of the privilege. As Chief Justice Marshall observed in United States v. Burr, 25 F. Cas. 38, 40 (No. 14,692e) (CC Va. 1807): "If, in such a case, he say upon his oath that his answer would incriminate himself, the court can demand no other testimony of the fact."

Our holding in Malloy v. Hogan, 378 U.S. 1 (1964), that the privilege applies to the States through the Fourteenth Amendment, determined that the right to remain silent is itself a liberty interest protected by that Amendment. We explained that "[t]he Fourteenth Amendment secures against state invasion the same privilege that the Fifth Amendment guarantees against federal infringement—the right of a person to remain silent unless he chooses to speak in the unfettered exercise of his own will, and to suffer no penalty . . . *58 for such silence." Id., at 8 (emphasis added). Since Malloy, we have construed the text to prohibit not only direct orders to testify, but also indirect compulsion effected by comments on a defendant's refusal to take the stand, Griffin v. California, 380 U.S. 609, 613-614 (1965), and we have recognized that compulsion can be presumed from the circumstances surrounding custodial interrogation, see Dickerson v. United States, 530 U.S. 428, 435 (2000) ("[T]he coercion inherent in custodial interrogation blurs the line between voluntary and involuntary statements, and thus heightens the risk that an individual will not be `accorded his privilege under the Fifth Amendment . . . not to be compelled to incriminate himself' ") (quoting Miranda v. Arizona, 384 U.S. 436, 439 (1966)). Without requiring the deprivation of any other liberty interest, we have found prohibited compulsion in the threatened loss of the right to participate in political associations, Lefkowitz v. Cunningham, 431 U.S. 801 (1977), forfeiture of government contracts, Turley, 414 U. S., at 82, loss of employment, Uniformed Sanitation Men Assn., Inc. v. Commissioner of Sanitation of City of New York, 392 U.S. 280 (1968), and disbarment, Spevack v. Klein, 385 U.S. 511, 516 (1967). None of our opinions contains any suggestion that compulsion should have a different meaning in the prison context. Nor is there any support in our Fifth Amendment jurisprudence for the proposition that nothing short of losing one's livelihood is sufficient to constitute compulsion. Accord, Turley, 414 U. S., at 83.

The plurality's suggestion that our decision in Meachum v. Fano, 427 U.S. 215 (1976), supports a novel interpretation of the Fifth Amendment, see ante, at 39, is inconsistent with the central rationale of that case. In Meachum, a group of prison inmates urged the Court to hold that the Due Process Clause entitled them to a hearing prior to their transfer to a substantially less favorable facility. Relying on the groundbreaking decisions in Morrissey v. Brewer, 408 U.S. 471 (1972), and Wolff v. McDonnell, 418 U.S. 539 (1974), *59 which had rejected the once-prevailing view that a prison inmate had no more rights than a "slave of the State,"^[5] the prisoners sought to extend those holdings to require judicial review of "any substantial deprivation imposed by prison authorities." The Court recognized that after Wolff and its progeny, convicted felons retain "a variety of important rights that the courts must be alert to protect." Although Meachum refused to expand the constitutional rights of inmates, we did not narrow the protection of any established right. Indeed, Justice White explicitly limited the holding to prison conditions that "do not otherwise violate the Constitution," 427 U.S., at 224.^[6]

Not a word in our discussion of the privilege in Ohio Adult Parole Authority v. Woodard, 523 U.S. 272 (1998), ante, at 43, requires a heightened showing of compulsion in the prison context to establish a Fifth Amendment violation. That case is wholly unlike this one because Woodard was not ordered to incriminate himself and was not punished for refusing to do so. He challenged Ohio's clemency procedures, arguing, inter alia, that an interview with members of the clemency board offered to inmates one week before their clemency hearing presented him with a Hobson's choice that violated the privilege against self-incrimination. He could either take advantage of the interview and risk incriminating himself, or decline the interview, in which case the clemency board might draw adverse inferences from his decision not to testify. We concluded that the prisoner who was offered "a voluntary interview" is in the same position as *60 any defendant faced with the option of either testifying or accepting the risk that adverse inferences may be drawn from his silence. 523 U. S.,at 286.

Respondent was directly ordered by prison authorities to participate in a program that requires incriminating disclosures, whereas no one ordered Woodard to do anything. Like a direct judicial order to answer questions in the courtroom, an order from the State to participate in the SATP is inherently coercive. Cf. Turley, 414 U. S., at 82 ("The waiver sought by the State, under threat of loss of contracts, would have been no less compelled than a direct request for the testimony without resort to the waiver"). Moreover, the penalty for refusing to participate in the SATP is automatic. Instead of conjecture and speculation about the indirect consequences that may flow from a decision to remain silent, we can be sure that defiance of a direct order carries with it the stigma of being a lawbreaker or a problem inmate, as well as other specified penalties. The penalty involved in this case is a mandated official response to the assertion of the privilege.

In Baxter v. Palmigiano, 425 U.S. 308 (1976), ante, at 42-43, we held that a prison disciplinary proceeding did not violate the privilege, in part, because the State had not "insisted [nor] asked that Palmigiano waive his Fifth Amendment privilege," and it was "undisputed that an inmate's silence in and of itself [was] insufficient to support an adverse decision by the Disciplinary Board." 425 U.S., at 317-318. We distinguished the "penalty cases," Garrity v. New Jersey, 385 U.S. 493 (1967), and Turley, not because they involved civilians as opposed to prisoners, as the plurality assumes, ante, at 40, but because in those cases the "refusal to submit to interrogation and to waive the Fifth Amendment privilege, standing alone and without regard to other evidence, resulted in loss of employment or opportunity to contract with the State," whereas Palmigiano's silence "was given no more evidentiary value than was warranted *61 by the facts surrounding his case." 425 U.S., at 318 (emphasis added). And, in a subsequent "penalty" case, we distinguished Baxter on the ground that refusing to incriminate oneself "was only one of a number of factors to be considered by the finder of fact in assessing a penalty, and was given no more probative value than the facts of the case warranted," while in Cunningham "refusal to waive the Fifth Amendment privilege [led] automatically and without more to imposition of sanctions." 431 U.S., at 808, n. 5.

Similarly, in Minnesota v. Murphy, 465 U. S., at 438, 439, while "the State could not constitutionally carry out a threat to revoke probation for the legitimate exercise of the Fifth Amendment privilege," because revocation was not automatic under the Minnesota statute, we concluded that "Murphy could not reasonably have feared that the assertion of the privilege would have led to revocation."^[7] These decisions recognized that there is an appreciable difference between an official sanction for disobeying a direct order and a mere risk of adverse consequences stemming from a voluntary choice. The distinction is not a novel one, nor is it simply offered to "justify departing from this Court's precedents," ante, at 44. Rather it is a distinction that we have drawn throughout our cases; therefore, it is the plurality's *62 disregard for both factors that represents an unjustified departure. Unlike Woodard, Murphy, and Baxter, respondent cannot invoke his Fifth Amendment rights and then gamble on whether the Department will revoke his Level III status; the punishment is mandatory. The fact that this case involves a prison inmate, as did Woodard and Baxter, is not enough to render those decisions controlling authority. Since we have already said inmates do not forfeit their Fifth Amendment rights at the jailhouse gate, Murphy, 465 U. S., at 426, the plurality must point to something beyond respondent's status as a prisoner to justify its departure from our precedent.

II

The plurality and Justice O'Connor hold that the consequences stemming from respondent's invocation of the privilege are not serious enough to constitute compulsion. The threat of transfer to Level I and a maximum-security unit is not sufficiently coercive in their view—either because the consequence is not really a penalty, just the loss of a benefit, or because it is a penalty, but an insignificant one. I strongly disagree.

It took respondent several years to acquire the status that he occupied in 1994 when he was ordered to participate in the SATP. Because of the nature of his convictions, in 1983 the Department initially placed him in a maximum-security classification. Not until 1989 did the Department change his "security classification to `medium by exception' because of his good behavior." Lile v. Simmons, 23 Kan. App. 2d 1, 2, 929 P.2d 171, 172 (1996). Thus, the sanction at issue threatens to deprive respondent of a status in the prison community that it took him six years to earn and which he had successfully maintained for five more years when he was ordered to incriminate himself. Moreover, abruptly "busting" his custody back to Level I, App. 94, would impose the same stigma on him as would a disciplinary conviction for any of the most serious offenses described in petitioners' formal *63 statement of Internal Management Policy and Procedure (IMPP). As the District Court found, the sanctions imposed on respondent "mirror the consequences imposed for serious disciplinary infractions." 24 F. Supp. 2d 1152, 1155 (Kan. 1998). This same loss of privileges is considered serious enough by prison authorities that it is used as punishment for theft, drug abuse, assault, and possession of dangerous contraband.^[8]

The punitive consequences of the discipline include not only the dignitary and reputational harms flowing from the transfer, but a serious loss of tangible privileges as well. Because he refused to participate in the SATP, respondent's visitation rights will be restricted. He will be able to earn only $0.60 per day, as compared to Level III inmates, who can potentially earn minimum wage. His access to prison organizations and activities will be limited. He will no longer be able to send his family more than $30 per pay period. He will be prohibited from spending more than $20 per payroll period at the canteen, rather than the $140 he could spend at Level III, and he will be restricted in what property he can keep in his cell. App. 27-28. In addition, because he will be transferred to a maximum-security unit, respondent will be forced to share a cell with three other *64 inmates rather than one, and his movement outside the cell will be substantially curtailed. Id., at 73, 83. The District Court found that the maximum-security unit is "a more dangerous environment occupied by more serious offenders." 24 F. Supp. 2d, at 1155.^[9] Perhaps most importantly, respondent will no longer be able to earn his way back up to Level III status through good behavior during the remainder of his sentence. App. 17 ("To complete Level I, an inmate must . . .demonstrate a willingness to participate in recommended programs and/or work assignments for a full review cycle").

The plurality's glib attempt to characterize these consequences as a loss of potential benefits rather than a penalty is wholly unpersuasive. The threatened transfer to Level I and to a maximum-security unit represents a significant, adverse change from the status quo. Respondent achieved his medium-security status after six years of good behavior and maintained that status during five more years. During that time, an inmate unquestionably develops settled expectations regarding the conditions of his confinement. These conditions then form the baseline against which any change must be measured, and rescinding them now surely constitutes punishment.

Paying attention to the baseline is not just "superficially appealing," ante, at 46. We have recognized that the government *65 can extend a benefit in exchange for incriminating statements, see Woodard, 523 U. S., at 288 ("[T]his pressure to speak in the hope of improving [one's] chance of being granted clemency does not make the interview compelled"), but cannot threaten to take away privileges as the cost of invoking Fifth Amendment rights, see, e. g., Turley, 414 U. S., at 82; Spevack, 385 U. S., at 516. Based on this distinction, nothing that I say in this dissent calls into question the constitutionality of downward adjustments for acceptance of responsibility under the United States Sentencing Guidelines, ante, at 47. Although such a reduction in sentence creates a powerful incentive for defendants to confess, it completely avoids the constitutional issue that would be presented if the Guidelines operated like the scheme here and authorized an upward adjustment whenever a defendant refused to accept responsibility. Similarly, taking into account an attorney's acceptance of responsibility or contrition in deciding whether to reinstate his membership to the bar of this Court, see ante, at 40, is obviously different from disbarring an attorney for invoking his privilege. By obscuring the distinction between penalties and incentives, it is the plurality that calls into question both the Guidelines and plea bargaining. See Corbitt v. New Jersey, 439 U.S. 212, 22-224 (1978) ("Nor does this record indicate that he was being punished for exercising a constitutional right. . . . [H]omicide defendants who are willing to plead non vult may be treated more leniently than those who go to trial, but withholding the possibility of leniency from the latter cannot be equated with impermissible punishment as long as our cases sustaining plea bargaining remain undisturbed").^[10]

*66 Even if the change in respondent's status could properly be characterized as a loss of benefits to which he had no entitlement, the question at hand is not whether the Department could have refused to extend those benefits in the first place, but rather whether revoking them at this point constitutes a penalty for asserting the Fifth Amendment privilege. See Perry v. Sindermann, 408 U.S. 593, 597 (1972). The plurality contends that the transfer from medium to maximum security and the associated loss of Level III status is not intended to punish prisoners for asserting their Fifth Amendment rights, but rather is merely incidental to the prison's legitimate interest in making room for participants *67 in the program. Ante, at 38. Of course, the Department could still house participants together without moving those who refuse to participate to more restrictive conditions of confinement and taking away their privileges. Moreover, petitioners have not alleged that respondent is taking up a bed in a unit devoted to the SATP; therefore, all the Department would have to do is allow respondent to stay in his current medium-security cell. If need be, the Department could always transfer respondent to another mediumsecurity unit. Given the absence of evidence in the record that the Department has a shortage of medium-security beds, or even that there is a separate unit devoted to participants in the SATP, the only plausible explanation for the transfer to maximum security and loss of Level III status is that it serves as punishment for refusing to participate in the program.

Justice O'Connor recognizes that the transfer is a penalty, but finds insufficient coercion because the "changes in [respondent's] living conditions seem to [her] minor." Ante, at 51 (opinion concurring in judgment). The coerciveness of the penalty in this case must be measured not by comparing the quality of life in a prison environment with that in a free society, but rather by the contrast between the favored and disfavored classes of prisoners. It is obviously impossible to measure precisely the significance of the difference between being housed in a four-person, maximum-security cell in the most dangerous area of the prison, on the one hand, and having a key to one's own room, the right to take a shower, and the ability to move freely within adjacent areas during certain hours, on the other—or to fully appreciate the importance of visitation privileges, being able to send more than $30 per pay period to family, having access to the yard for exercise, and the opportunity to participate in group activities. What is perfectly clear, however, is that it is the aggregate effect of those penalties that creates compulsion. Nor is it coincidental that petitioners have selected this same *68 group of sanctions as the punishment to be imposed for the most serious violations of prison rules. Considering these consequences as a whole and comparing the Department's treatment of respondent to the rest of the prison population, it is perfectly clear that the penalty imposed is "constitutionally indistinguishable from the coercive provisions we struck down in Gardner, Sanitation Men, and Turley. " Cunningham, 431 U. S., at 807.^[11]

III

The SATP clearly serves legitimate therapeutic purposes. The goal of the program is to rehabilitate sex offenders, and the requirement that participants complete admission of responsibility and sexual history forms may well be an important component of that process. Mental health professionals seem to agree that accepting responsibility for past sexual misconduct is often essential to successful treatment, and that treatment programs can reduce the risk of recidivism by sex offenders. See Winn, Strategic and Systematic Management of Denial in Cognitive/Behavioral Treatment of Sexual Offenders, 8 Sexual Abuse: J. Research and Treatment 25, 26-27 (1996).

The program's laudable goals, however, do not justify reduced constitutional protection for those ordered to participate. "We have already rejected the notion that citizens may be forced to incriminate themselves because it serves a governmental need." Cunningham, 431 U. S., at 808. *69 The benefits of obtaining confessions from sex offenders may be substantial, but "claims of overriding interests are not unusual in Fifth Amendment litigation," and until today at least "they have not fared well." Turley, 414 U. S., at 78. The State's interests in law enforcement and rehabilitation are present in every criminal case. If those interests were sufficient to justify impinging on prisoners' Fifth Amendment right, inmates would soon have no privilege left to invoke.

The plurality's willingness to sacrifice prisoners' Fifth Amendment rights is also unwarranted because available alternatives would allow the State to achieve the same objectives without impinging on inmates' privilege. Turner v. Safley, 482 U.S. 78, 93 (1987). The most obvious alternative is to grant participants use immunity. See Murphy, 465 U. S., at 436, n. 7 ("[A] State may validly insist on answers to even incriminating questions . . . as long as it recognizes that the required answers may not be used in a criminal proceeding and thus eliminates the threat of incrimination"); Baxter, 425 U. S., at 318 ("Had the State desired Palmigiano's testimony over his Fifth Amendment objection, we can but assume that it would have extended whatever use immunity is required by the Federal Constitution"). Petitioners have not provided any evidence that the program's therapeutic aims could not be served equally well by granting use immunity. Participants would still obtain all the therapeutic benefits of accepting responsibility and admitting past misconduct; they simply would not incriminate themselves in the process. At least one State already offers such protection, see Ky. Rev. Stat. Ann. § 197.440 (West 2001) ("Communications made in the application for or in the course of a sexual offender's diagnosis and treatment . . . shall be privileged from disclosure in any civil or criminal proceeding"), and there is no indication that its choice is incompatible with rehabilitation. In fact, the program's rehabilitative goals would likely be furthered by ensuring free and open discussion *70 without the threat of prosecution looming over participants' therapy sessions.

The plurality contends that requiring immunity will undermine the therapeutic goals of the program because once "inmates know society will not punish them for their past offenses, they may be left with the false impression that society does not consider those crimes to be serious ones." Ante, at 34. See also Brief for 18 States as Amici Curiae 11 ("By subjecting offenders to prosecution for newly revealed offenses, and by adhering to its chosen policy of mandatory reporting for cases of suspected child sexual abuse, Kansas reinforces the sensible notion that wrongdoing carries consequences"). The idea that an inmate who is confined to prison for almost 20 years for an offense could be left with the impression that his crimes are not serious or that wrongdoing does not carry consequences is absurd. Moreover, the argument starts from a false premise. Granting use immunity does not preclude prosecution; it merely prevents the State from using an inmate's own words, and the fruits thereof, against him in a subsequent prosecution. New Jersey v. Portash, 440 U.S. 450, 457-458 (1979). The plurality's concern might be justified if the State were required to grant transactional immunity, but we have made clear since Kastigar that use immunity is sufficient to alleviate a potential Fifth Amendment violation, 406 U.S., at 453. Nor is a State required to grant use immunity in order to have a sex offender treatment program that involves admission of responsibility.

Alternatively, the State could continue to pursue its rehabilitative goals without violating participants' Fifth Amendment rights by offering inmates a voluntary program. The United States points out that an inmate's participation in the sexual offender treatment program operated by the Federal Bureau of Prisons is entirely voluntary. "No loss of institutional privileges flows from an inmate's decision not to participate *71 in the program."^[12] If an inmate chooses to participate in the federal program, he will be transferred from his "parent facility" to a "more desirable" prison, but if he refuses to participate in the first place, as respondent attempted to do, he suffers no negative consequences. Tr. of Oral Arg. 21-22. Although the inmates in the federal program are not granted use immunity, they are not compelled to participate. Indeed, there is reason to believe successful rehabilitation is more likely for voluntary participants than for those who are compelled to accept treatment. See Abel, Mittelman, Becker, Rathner, & Rouleau, Predicting Child Molesters' Response to Treatment, 528 Annals N. Y. Acad. of Sciences 223 (1988) (finding that greater perceived pressure to participate in treatment is strongly correlated with the dropout rate).

Through its treatment program, Kansas seeks to achieve the admirable goal of reducing recidivism among sex offenders. In the process, however, the State demands an impermissible and unwarranted sacrifice from the participants. No matter what the goal, inmates should not be compelled to forfeit the privilege against self-incrimination simply because the ends are legitimate or because they have been convicted of sex offenses. Particularly in a case like this one, in which respondent has protested his innocence all along and is being compelled to confess to a crime that he still insists he did not commit, we ought to ask ourselves—what if this is one of those rare cases in which the jury made a *72 mistake and he is actually innocent? And in answering that question, we should consider that even members of the Star Chamber thought they were pursuing righteous ends.

I respectfully dissent.

NOTES

[*] A brief of amici curiae urging reversal was filed for the State of Ohio et al. by Betty D. Montgomery, Attorney General of Ohio, David M. Gormley, State Solicitor, Todd R. Marti, Assistant Solicitor, Mike McGrath, Attorney General of Montana, Jenifer Anders, Assistant Attorney General, and by the Attorneys General for their respective States as follows: Bill Pryor of Alabama, Janet Napolitano of Arizona, Ken Salazar of Colorado, M. Jane Brady of Delaware, Robert A. Butterworth of Florida, Steve Carter of Indiana, J. Joseph Curran, Jr., of Maryland, Thomas F. Reilly of Massachusetts, Don Stenberg of Nebraska, Frankie Sue Del Papa of Nevada, Patricia A. Madrid of New Mexico, Charles M. Condon of South Carolina, Mark L. Shurtleff of Utah, Randolph A. Beales of Virginia, Christine O. Gregoire of Washington, and Gay Woodhouse of Wyoming.

[1] As a participant in the Sexual Abuse Treatment Program (SATP), respondent would be required to sign an "Admission of Responsibility" form setting forth the details of the offense for which he was convicted. Because he had testified at trial that his sexual intercourse with the victim before driving her back to her car was consensual, the District Court found that a written admission on this form would subject respondent to a possible charge of perjury. 24 F. Supp. 2d 1152, 1157 (Kan. 1998). In addition, the SATP requires participants to "generate a written sexual history which includes all prior sexual activities, regardless of whether such activities constitute uncharged criminal offenses." Id., at 1155. The District Court found that the form "clearly seeks information that could incriminate the prisoner and subject him to further criminal charges." Id., at 1157.

[2] Petitioners relied on two cases, Fisher v. United States, 425 U.S. 391 (1976), and United States v. Washington, 431 U.S. 181, 187-188 (1977). In Fisher, we held that the privilege does not permit the target of a criminal investigation to prevent his lawyer from answering a subpoena to produce incriminating documents. We reached that conclusion because the person asserting the privilege was not the one being compelled. In Washington, cited ante, at 36, a grand jury witness voluntarily answered questions after being advised of the privilege, though not of the fact that he was a potential defendant in danger of being indicted. In neither case did the witness assert the privilege against incriminating himself.

[3] The origins and evolution of the privilege have received significant scholarly attention and debate in recent years. See, e. g., Hazlett, Nineteenth Century Origins of the Fifth Amendment Privilege Against SelfIncrimination, 42 Am. J. Legal Hist. 235 (1998); Amar & Lettow, Fifth Amendment First Principles: The Self-Incrimination Clause, 93 Mich. L. Rev. 857 (1995). The historical account is complicated by the fact that before Boyd v. United States, 116 U.S. 616 (1886), the privilege was treated as a common-law evidentiary doctrine separate from the Fifth Amendment. During that time, the privilege was also subsumed within general discussions of the voluntariness of confessions.

[4] Alschuler, A Peculiar Privilege in Historical Perspective, in The Privilege Against Self-Incrimination 181, 192-193 (R. Helmholz et al. eds. 1997) (discussing historical sources which indicate that the "privilege prohibited (1) incriminating interrogation under oath, (2)torture, and (3) probably other forms of coercive interrogation such as threats of future punishment and promises of leniency" (footnotes omitted)).

[5] See Meachum v. Fano, 427 U.S. 215, 231 (1976) (Stevens, J., dissenting).

[6] In his opinion for the Court in the companion case, Montanye v. Haymes, 427 U.S. 236, 242 (1976), Justice White reiterated this point: "As long as the conditions or degree of confinement to which the prisoner is subjected are within the sentence imposed upon him and [are] not otherwise violative of the Constitution, the Due Process Clause does not in itself subject an inmate's treatment by prison authorities to judicial oversight."

[7] The plurality is quite wrong to rely on Murphy for the proposition that an individual is not compelled to incriminate himself when faced with the threat of return to prison. Ante, at 43. In Murphy, we did not have occasion to decide whether such a threat constituted compulsion because we held that "since Murphy revealed incriminating information instead of timely asserting his Fifth Amendment privilege, his disclosures were not compelled incriminations." 465 U. S.,at 440. As we explained, "a witness confronted with questions that the government should reasonably expect to elicit incriminating evidence ordinarily must assert the privilege rather than answer if he desires not to incriminate himself. .. .But if he chooses to answer, his choice is considered to be voluntary since he was free to claim the privilege and would suffer no penalty as the result of his decision to do so." Id., at 429. In contrast to Murphy, respondent has consistently asserted his Fifth Amendment privilege.

[8] IMPP 11-101 provides that an inmate "shall be automatically reduced to Level I for any of the following: (1) Termination from a work or program assignment for cause;(2) Refusal to participate in recommended programs at the time of placement; (3) Offenses committed in which a felony charge is filed with the district or county prosecutor; (4) Disciplinary convictions for:(a)Theft; (b) Being in a condition of drunkenness, intoxication, or a state of altered consciousness; (c) Use of stimulants, sedatives, unauthorized drugs, or narcotics, or the misuse, or hoarding of authorized or prescribed medication; (d) Sodomy, aggravated sodomy, or aggravated sexual act; (e) Riot or incitement to riot; (f) Arson; (g) Assault; (h) Battery; (i) Inmate Activity (limitations); (j) Sexual Activity; (k) Interference with Restraints;(l) Relationships with Staff; (m) Work Performance; or (n) Dangerous Contraband." App. 19-20 (citations omitted).

[9] Respondent attested to the fact that in his experience maximum security "is a very hostile, intimidating environment because most of the inmates in maximum tend to have longer sentences and are convicted of more serious crimes, and, as a consequence, care less how they act or treat others." Id., at 41-42. He explained that in the maximum-security unit "there is far more gang activity,""reported and unreported rapes and assaults of inmates are far more prevalent," and "sex offenders .. .are seen as targets for rape and physical and mental assault[s]," whereas in medium security, "because the inmates want to maintain their medium security status, they are less prone to breaking prison rules or acting violently." Id., at 42-43.

[10] The plurality quotes a footnote in Roberts v. United States, 445 U.S. 552 (1980), for the proposition that a principled distinction cannot be drawn between enhancing punishment and denying leniency, ante, at 46. This quote is misleading because, as in Minnesota v. Murphy, 465 U.S. 420 (1984), see n. 7,supra, Roberts failed to assert his privilege against self-incrimination, and we reiterated that the privilege is not selfexecuting, 445 U.S., at 559. Furthermore, the passage quoted by the plurality, id., at 557, n. 4, was in reference to Roberts' claim that the sentencing judge could not consider his refusal to incriminate a co-conspirator in deciding whether to impose his sentences consecutively. In that context, the privilege is not implicated and compulsion is not constitutionally significant. While it is true that in some cases the line between enhancing punishment and refusing leniency may be difficult to draw, that does not mean the distinction is irrelevant for Fifth Amendment purposes.

It is curious that the plurality asserts the impracticality of drawing such a distinction, given that in this case a majority of the Court agrees that it is perfectly clear the consequences facing respondent represent a burden, rather than the denial of a benefit. Ante, at 53-54 (O'Connor, J., concurring in judgment). Our cases reveal that it is not only possible, but necessary to draw the distinction. For even Bordenkircher v. Hayes, 434 U.S. 357 (1978), conditioned its entire analysis of plea bargaining on the assumption that the defendant had been charged with the greater offense prior to plea bargaining and, therefore, faced the denial of leniency rather than an enhanced penalty. Id., at 360-361 ("While the prosecutor did not actually obtain the recidivist indictment until after the plea conferences had ended, his intention to do so was clearly expressed at the outset of plea negotiations. . . . This is not a situation, therefore, where the prosecutor without notice brought an additional and more serious charge after plea negotiations relating only to the original indictment had ended with the defendant's insistence on pleading not guilty. As a practical matter, in short, this case would be no different if the grand jury had indicted [the defendant] as a recidivist from the outset, and the prosecutor had offered to drop that charge as part of the plea bargain").

[11] Justice O'Connor would distinguish these cases because the penalty involved the loss of one's livelihood, whereas here respondent will be housed, clothed, and fed regardless of whether he is in maximum or medium security. We rejected a similar argument in Turley, when we refused to distinguish Gardner v. Broderick, 392 U.S. 273 (1968), and Uniformed Sanitation Men Assn., Inc. v. Commissioner of Sanitation of City of New York, 392 U.S. 280 (1968), based on the difference between losing one's job and losing the ability to obtain government contracts. 414 U.S., at 83. We concluded that there was no "difference of constitutional magnitude between the threat of job loss to an employee of the State, and a threat of loss of contracts to a contractor." Ibid.

[12] Brief for United States as Amicus Curiae 27. Because of this material difference between the Kansas and federal programs, recognizing the compulsion in this case would not cast any doubt on the validity of voluntary programs. The plurality asserts that "the federal program is different from Kansas' SATP only in that it does not require inmates to sacrifice privileges besides housing as a consequence of nonparticipation." Ante, at 45 (emphasis added). This statement is inaccurate because, as the quote in the text reveals, no loss of privileges follows from the decision not to participate in the federal program.