In 1983 and thereafter the Florida Legislature enacted a series of statutes authorizing the Department of Corrections to award early release credits to prison inmates when the population of the state prison system exceeded predetermined levels. The question presented by this case is whether a 1992 statute canceling such credits for certain classes of offenders after they had been awarded-indeed, after they had resulted in the prisoners' release from custody-violates the Ex Post Facto Clause of the Federal Constitution.

* In 1986 petitioner pleaded nolo contendere to a charge of attempted murder and received a sentence of 22 years (8,030 days) in prison. In 1992 the Florida Department of Corrections released him from prison based on its determination that he had accumulated five different types of early release credits totaling 5,668 days.1 Of that total, 1,860 days were "provisional credits'' awarded as a result of prison overcrowding. Shortly after petitioner's release, the state attorney general issued an opinion interpreting a 1992 statute as having retroactively canceled all provisional credits awarded to inmates convicted of murder or attempted murder. Petitioner was therefore rearrested and returned to custody. His new release date was set for May 19, 1998.

In 1994 petitioner filed a petition for a writ of habeas corpus alleging that the retroactive cancellation of provisional credits violated the Ex Post Facto Clause. Relying on 11th Circuit2 and Florida3 precedent holding that the revocation of provisional credits did not violate the Ex Post Facto Clause because their sole purpose was to alleviate prison overcrowding, the Magistrate Judge recommended dismissal of the petition. The District Court adopted that recommendation, dismissed the petition, and denied a certificate of probable cause. The Court of Appeals for the 11th Circuit also denied a certificate of probable cause in an unpublished order. Because the Court of Appeals for the 10th Circuit reached a different conclusion on similar facts, Arnold v. Cody, 951 F.2d 280 (1991), we granted certiorari to resolve the conflict. 517 U.S. ____, 116 S.Ct. 1671, 134 L.Ed.2d 775 (1996).4

Motivated largely by the overcrowded condition of the entire Florida prison system,5 in 1983 the State Legislature enacted the Correctional Reform Act of 1983, a comprehensive revision of the State's sentencing laws.6 The Act authorized generous awards of early release credits including "basic gain-time'' at the rate of 10 days for each month, "up to 20 days of incentive gain time, which shall be credited and applied monthly,'' and additional deductions of "meritorious gain-time of from 1 to 60 days.'' See 1983 Fla. Laws, ch. 83-131 §8.7 The Act also created an emergency procedure to be followed "whenever the population of the state correctional system exceeds 98 percent of the lawful capacity of the system for males or females, or both.'' §5(1).8 When such an emergency was declared, "the sentences of all inmates in the system who are eligible to earn gain-time shall be reduced by the credit of up to 30 days gain-time in 5-day increments as may be necessary to reduce the inmate population to 97 percent of lawful capacity.'' §5(2).

In the ensuing years, the Florida Legislature modified the overcrowding gain-time system. In 1987 the legislature raised the threshold for awarding emergency release credits from 98% to 99% of capacity. At the same time, the legislature authorized a new form of overcrowding credit, administrative gain-time, with a 98% threshold, which authorized up to a maximum of 60 days additional gain-time to inmates already earning incentive gain-time. Inmates serving sentences for certain offenses were ineligible for the awards. In 1988 the legislature repealed the administrative gain-time provision, and replaced it with a provisional credits system.9 The language of the provisional credits statute was virtually identical to that of the administrative gain-time statute-it also authorized up to 60 days of gain-time but was triggered when the inmate population reached 97.5% of capacity. In addition, the legislature expanded the list of offenders who were ineligible for the awards.

Having received overcrowding gain-time under the administrative gain-time and provisional credits statutes, as well as basic and incentive gain-time, petitioner was released from prison in 1992. That same year, the legislature canceled provisional overcrowding credits for certain classes of inmates, including those convicted of attempted murder.10 As a result of that action, credits for 2,789 inmates who were still in custody were canceled, and rearrest warrants were issued for 164 offenders who had been released.11 Petitioner was in the latter class.

Respondents contend that the cancellation of petitioner's provisional credits did not violate the Ex Post Facto Clause for two reasons: (1) because the credits had been issued as part of administrative procedures designed to alleviate overcrowding, they were not an integral part of petitioner's punishment; and (2) in petitioner's case, the specific overcrowding credits had been awarded pursuant to statutes enacted after the date of his offense rather than pursuant to the 1983 statute. We consider the arguments separately.

The presumption against the retroactive application of new laws is an essential thread in the mantle of protection that the law affords the individual citizen. That presumption "is deeply rooted in our jurisprudence, and embodies a legal doctrine centuries older than our Republic.'' Landgraf v. USI Film Products, 511 U.S. 244, 265, 114 S.Ct. 1483, 1497, 128 L.Ed.2d 229 (1994). This doctrine finds expression in several provisions of our Constitution.12 The specific prohibition on ex post facto laws is only one aspect of the broader constitutional protection against arbitrary changes in the law. In both the civil and the criminal context, the Constitution places limits on the sovereign's ability to use its law-making power to modify bargains it has made with its subjects. The basic principle is one that protects not only the rich and the powerful, United States v. Winstar Corp., 518 U.S. ____, 116 S.Ct. 2432, 135 L.Ed.2d 964 (1996), but also the indigent defendant engaged in negotiations that may lead to an acknowledgment of guilt and a suitable punishment.

Article I, §10 of the Federal Constitution provides that " [n]o State shall . . . pass any . . . ex post facto Law.'' In his opinion for the Court in Beazell v. Ohio, 269 U.S. 167, 46 S.Ct. 68, 70 L.Ed. 216 (1925), Justice Stone explained:

"The constitutional prohibition and the judicial interpretation of it rest upon the notion that laws, whatever their form, which purport to make innocent acts criminal after the event, or to aggravate an offense, are harsh and oppressive, and that the criminal quality attributable to an act, either by the legal definition of the offense or by the nature or amount of the punishment imposed for its commission, should not be altered by legislative enactment, after the fact, to the disadvantage of the accused.'' Id., at 170, 46 S.Ct., at 68-69.

The bulk of our ex post facto jurisprudence has involved claims that a law has inflicted a "a greater punishment, than the law annexed to the crime, when committed.'' Calder v. Bull, 3 Dall. 386, 390, 1 L.Ed. 648 (1798).13 We have explained that such laws implicate the central concerns of the Ex Post Facto Clause: "the lack of fair notice and governmental restraint when the legislature increases punishment beyond what was prescribed when the crime was consummated.'' Weaver v. Graham, 450 U.S. 24, 30, 101 S.Ct. 960, 965, 67 L.Ed.2d 17 (1981).

To fall within the ex post facto prohibition, a law must be retrospective-that is "it must apply to events occurring before its enactment''-and it "must disadvantage the offender affected by it'' id., at 29, 101 S.Ct., at 964, by altering the definition of criminal conduct or increasing the punishment for the crime, see Collins v. Youngblood, 497 U.S. 37, 50, 110 S.Ct. 2715, 2723, 111 L.Ed.2d 30 (1990). In this case the operation of the 1992 statute to effect the cancellation of overcrowding credits and the consequent reincarceration of petitioner was clearly retrospective. The narrow issue that we must decide is thus whether those consequences disadvantaged petitioner by increasing his punishment.

In arguing that the cancellation of overcrowding credits inflicts greater punishment, petitioner relies primarily on our decision in Weaver v. Graham, in which we considered whether retroactively decreasing the amount of gain-time awarded for an inmate's good behavior violated the Ex Post Facto Clause. In that case the petitioner had pleaded guilty to second-degree murder and had been sentenced to prison for 15 years. At the time of Weaver's plea, Florida law provided credits contingent on the good conduct of the prisoner of 5 days per month for the first two years of his sentence, 10 days per month for the third and fourth years, and 15 days per month thereafter. The law therefore provided him with an opportunity to be released after serving less than nine years of his sentence. In 1978 the Florida Legislature enacted a new formula for computing gain time; instead of 5, 10, and 15 days per month, it authorized only three, six, and nine days. The new statute did not withdraw any credits already awarded to Weaver, but by curtailing the availability of future credits it effectively postponed the date when he would become eligible for early release. Because the statute made the punishment for crimes committed before its enactment "more onerous,'' we unanimously concluded that it ran "afoul of the prohibition against ex post facto laws.'' Id., at 36, 101 S.Ct., at 968.

According to petitioner, although this case involves overcrowding credits, it is essentially like Weaver because the issuance of these credits was dependent on an inmate's good conduct. Respondents on the other hand submit that Weaver is not controlling because it was the overcrowded condition of the prison system, rather than the character of the prisoner's conduct, that gave rise to the award. In our view, both of these submissions place undue emphasis on the legislature's subjective intent in granting the credits rather than on the consequences of their revocation.

In arriving at our holding in Weaver, we relied not on the subjective motivation of the legislature in enacting the gain-time credits, but rather on whether objectively the new statute "lengthen[ed] the period that someone in petitioner's position must spend in prison.'' Id., at 33, 101 S.Ct., at 967. Similarly, in this case, the fact that the generous gain-time provisions in Florida's 1983 statute were motivated more by the interest in avoiding overcrowding than by a desire to reward good behavior, is not relevant to the essential inquiry demanded by the Ex Post Facto Clause: whether the cancellation of 1,860 days of accumulated provisional credits had the effect of lengthening petitioner's period of incarceration.

In our post-Weaver cases, we have also considered whether the legislature's action lengthened the sentence without examining the purposes behind the original sentencing scheme. In Miller v. Florida, 482 U.S. 423, 107 S.Ct. 2446, 96 L.Ed.2d 351 (1987), we unanimously concluded that a revision in Florida's sentencing guidelines that went into effect between the date of petitioner's offense and the date of his conviction violated the Ex Post Facto Clause. Our determination that the new guideline was "more onerous than the prior law,'' id., at 431, 107 S.Ct., at 2452 (quoting Dobbert v. Florida, 432 U.S. 282, 294, 97 S.Ct. 2290, 2298-2299, 53 L.Ed.2d 344 (1977)), rested entirely on an objective appraisal of the impact of the change on the length of the offender's presumptive sentence. 482 U.S., at 431, 107 S.Ct., at 2452. —("Looking only at the change in primary offense points, the revised guidelines law clearly disadvantages petitioner and similarly situated defendants'').

In California Dept. of Corrections v. Morales, 514 U.S. ____, 115 S.Ct. 1597, 131 L.Ed.2d 588 (1995), we also relied entirely on objective considerations to support our conclusion that an amendment to California's parole procedures that decreased the frequency of parole hearings for certain offenders had not made any "change in the "quantum of punishment,''' id., at ____, 115 S.Ct., at 1602. The amendment at issue in Morales allowed the Parole Board, after holding an initial parole hearing, to defer for up to three years subsequent parole suitability hearings for prisoners convicted of multiple murders if the Board found that it was unreasonable to expect that parole would be granted at a hearing during the subsequent years. We stated that the relevant inquiry is whether the "change alters the definition of criminal conduct or increases the penalty by which a crime is punishable.'' Id., at ____, n. 3, 115 S.Ct., at 1602, n. 3.14 After making that inquiry, we found that "there is no reason to conclude that the amendment will have any effect on any prisoner's actual term of confinement.'' Id., at ____, 115 S.Ct., at 1604. Our holding rested squarely on the conclusion that "a prisoner's ultimate date of release would be entirely unaffected by the change in the timing of suitability hearings.'' Id., at ____, 115 S.Ct., at 1605. Although we held that "speculative and attenuated possibilities'' of increasing the measure of punishment do not implicate the Ex Post Facto Clause, id., at ____, 115 S.Ct., at 1609, the bulk of our analysis focused on the effect of the law on the inmate's sentence.

We did not imply in Morales, as respondents contend, that the constitutionality of retroactive changes in the quantum of punishment depended on the purpose behind the parole sentencing system. The only mention of legislative purpose in Morales, was in the following passage:

"In contrast to the laws at issue in Lindsey [v. Washington, 301 U.S. 397, 57 S.Ct. 797, 81 L.Ed. 1182 (1937)] Weaver, and Miller (which had the purpose and effect of enhancing the range of available prison terms, see Miller, supra, at 433-434 [107 S.Ct., at 2452-2453]), the evident focus of the California amendment was merely ""to relieve the [Board] from the costly and time-consuming responsibility of scheduling parole hearings'''' for prisoners who have no reasonable chance of being released. In re Jackson, 39 Cal.3d 464, 473 [216 Cal.Rptr. 760, 765-766] 703 P.2d 100, 106 (1985) (quoting legislative history).'' 514 U.S., at ____, 115 S.Ct., at 1602.

Thus, we concluded, the change at issue had neither the purpose nor the effect of increasing the quantum of punishment. Whether such a purpose alone would be a sufficient basis for concluding that a law violated the Ex Post Facto Clause when it actually had no such effect is a question the Court has never addressed. Moreover, in Morales our statements regarding purpose did not refer to the purpose behind the creation of the original sentencing scheme; they referred instead to the question whether, in changing that sentencing scheme, the legislature intended to lengthen the inmate's sentence. To the extent that any purpose might be relevant in this case, it would only be the purpose behind the legislature's 1992 enactment of the offense-based exclusion. Here, unlike in Morales, there is no evidence that the legislature's change in the sentencing scheme was merely to save time or money. Rather, it is quite obvious that the retrospective change was intended to prevent the early release of prisoners convicted of murder-related offenses who had accumulated overcrowding credits.15

Respondents also argue that the retroactive cancellation of overcrowding credits is permissible because overcrowding gain-time-unlike the incentive gain-time at issue in Weaver which is used to encourage good prison behavior and prisoner rehabilitation-"b[ears] no relationship to the original penalty assigned the crime or the actual penalty calculated under the sentencing guidelines.'' Brief for Respondent Mathis 20. To the extent that respondents' argument rests on the notion that overcrowding gain-time is not "in some technical sense part of the sentence,'' Weaver, 450 U.S., at 32, 101 S.Ct., at 966, this argument is foreclosed by our precedents. As we recognized in Weaver, retroactive alteration of parole or early release provisions, like the retroactive application of provisions that govern initial sentencing, implicates the Ex Post Facto Clause because such credits are "one determinant of petitioner's prison term . . . and . . . [the petitioner's] effective sentence is altered once this determinant is changed.'' Ibid. We explained in Weaver that the removal of such provisions can constitute an increase in punishment, because a "prisoner's eligibility for reduced imprisonment is a significant factor entering into both the defendant's decision to plea bargain and the judge's calculation of the sentence to be imposed.'' Ibid.

Respondents argue that this reasoning does not apply to overcrowding credits because, when petitioner pleaded nolo contendere, he could not reasonably have expected to receive any such credits. The State, after all, could have alleviated the overcrowding problem in various ways: it could have built more prisons; it could have paroled a large category of non-violent offenders; or it might have discontinued prosecution of some classes of victimless crimes. Respondents thus argue that the 1992 statute does not violate the Ex Post Facto Clause because, like the California Amendment at issue in Morales, it "create[d] only the most speculative and attenuated possibility of producing the prohibited effect of increasing the measure of punishment for covered crimes.'' 514 U.S., at ____, 115 S.Ct., at 1603.16 Given the fact that this petitioner was actually awarded 1,860 days of provisional credits and the fact that those credits were retroactively cancelled as a result of the 1992 amendment, we find this argument singularly unpersuasive. In this case, unlike in Morales, the actual course of events makes it unnecessary to speculate about what might have happened. The 1992 statute has unquestionably disadvantaged the petitioner because it resulted in his rearrest and prolonged his imprisonment. Unlike the California amendment at issue in Morales, the 1992 Florida statute did more than simply remove a mechanism that created an opportunity for early release for a class of prisoners whose release was unlikely; rather it made ineligible for early release a class of prisoners who were previously eligible-including some, like petitioner, who had actually been released.17

Although it does not appear that respondents advanced this argument in the papers filed in the District Court, the Court of Appeals, or in their brief in opposition to the petition for certiorari in this Court, they now argue that petitioner is not entitled to relief because his overcrowding credits were awarded pursuant to statutes enacted after the date of his offense rather than pursuant to the 1983 statute. We disagree.

The overcrowding statute in effect at the time of petitioner's crime was modified in subsequent years, but its basic elements remained the same: the statute provided for reductions in a prisoner's sentence when the population of the prison system exceeded a certain percentage of lawful capacity. At the time of petitioner's sentence in 1986, the emergency gain-time statute was in effect. Under that statute, when the prison population reached 98% of lawful capacity, the Secretary of the Department of Corrections was required to advise the Governor and, after receiving the Governor's verification of the capacity certification, to declare a state of emergency whereupon the sentences of all eligible inmates "shall be reduced by the credit of up to 30 days gain-time, in 5-day increments, as may be necessary to reduce the inmate population to 97 percent of lawful capacity.'' Fla. Stat. §944.598(2) (1983).18

FC21SMoreover, the attorney general's reading of the emergency gain-time statute would also limit the award of gain-time under the administrative gain-time and provisional credits statute. These statutes contain wording similar to the emergency gain-time statute, see Fla. Stat. §944.276(1) (1987), Fla. Stat. §944.277(1) (1989), yet the State has not interpreted the statutes to limit the award of gain-time to a total of 60 days.

The later statutes slightly modified the procedures outlined in the 1983 statute. The administrative gain-time statute enacted in 1987 (after petitioner's plea of nolo contendere) provided that the Secretary, after certification to the Governor, "may grant up to a maximum of 60 days administrative gain-time.'' Fla. Stat. §944.276(1) (1987). Unlike the emergency gain-time statute, the administrative gain-time statute made the issuance of gain-time discretionary, and it contained certain offense-based exclusions. The provisional credits provision was enacted to replace administrative gain time and is essentially the same, except that it provides for the issuance of gain-time when the prison reaches 97.5% of lawful capacity, rather than 98%. Fla. Stat. §944.277 (1988). See Griffin v. Singletary, 638 So.2d 500 (Fla.1994).

The changes in the series of statutes authorizing the award of overcrowding gain-time, do not affect petitioner's core ex post facto claim. Petitioner could have accumulated gain-time under the emergency gain-time provision in much the same manner as he did under the provisional credits statute. We recognize, however, that although the differences in the statutes did not affect petitioner's central entitlement to gain-time, they may have affected the precise amount of gain-time he received. Between 1988 and 1992, the provisional credits were authorized when the prison reached 97.5% capacity rather than 98% capacity as under the emergency gain-time statute. If the prison population did not exceed 98% of capacity between 1988 and 1992, and if petitioner received provisional credits during those years, there is force to the argument that the cancellation of that portion of the 1,860-day total did not violate the Ex Post Facto Clause. Because this point was not adequately developed earlier in the proceeding, and because it may not in any event affect petitioner's entitlement to release, we leave it open for further consideration on remand.

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

It is so ordered.

Justice THOMAS, with whom Justice SCALIA joins, concurring in part and concurring in the judgment.

I understand the Court's opinion to hold that retroactively cancelling petitioner's so-called "provisional credits'' after he has used them to gain his freedom, violates the Ex Post Facto Clause. This result naturally follows from our consistent view that the Clause is intended to prohibit laws that "retroactively alter the definition of crimes or increase the punishment for criminal acts.'' Collins v. Youngblood, 497 U.S. 37, 43, 110 S.Ct. 2715, 2719, 111 L.Ed.2d 30 (1990).

Whether a particular law retroactively increases a criminal punishment is often a close question. In California Dept. of Corrections v. Morales, 514 U.S. ____, 115 S.Ct. 1597, 131 L.Ed.2d 588 (1995), for example, respondent challenged a retroactive change to the frequency of parole hearings. Given that the retroactive change "create[d] only the most speculative and attenuated risk of increasing the measure of punishment attached to the covered crimes,'' we found no ex post facto violation. Morales, 514 U.S., at ____, 115 S.Ct., at 1605.

Unlike in Morales, the increase in petitioner's punishment here was neither "speculative'' nor "attenuated.'' Petitioner pleaded nolo contendre to a charge of attempted murder and was duly sentenced. During the period of his confinement, petitioner accumulated release credits under a state statute adopted in response to prison overcrowding. Those credits enabled petitioner to be freed from prison before his sentence (as originally imposed) had run. Shortly before petitioner secured his release, however, the Florida legislature enacted a statute preventing certain categories of offenders from taking advantage of the provisional credits. Although petitioner's offense placed him among the offenders denied the opportunity to acquire those particular credits, the statute was not applied retroactively. Petitioner was thus released. The State Attorney General subsequently issued an opinion giving the statute retroactive effect. The State thereafter rearrested petitioner and returned him to custody.

Under these narrow circumstances, I agree with the Court that the State's retroactive nullification of petitioner's previously accrued, and then used, release credits violates the Constitution's ban on ex post facto lawmaking. I do not, however, join the majority's discussion of Weaver v. Graham, 450 U.S. 24, 101 S.Ct. 960, 67 L.Ed.2d 17 (1981), which I find unnecessary to the resolution of this case. In Weaver, we considered whether a statute that merely altered the availability of "good conduct'' credits ran afoul of the Ex Post Facto Clause. Id., at 25, 101 S.Ct., at 962. The present case involves not merely an effect on the availability of future release credits, but the retroactive elimination of credits already earned and used. Accordingly, I concur in part and concur in the judgment.