Opinion of the Court JOHNSON v. UNITED STATES 529 U. S. ____ (2000) NOTICE: This opinion is subject to formal revision before publication in the preliminary print of the United States Reports. Readers are requested to notify the Reporter of Decisions, Supreme Court of the United States, Washington, D. C. 20543, of any typographical or other formal errors, in order that corrections may be made before the preliminary print goes to press. SUPREME COURT OF THE UNITED STATES No. 99–5153

CORNELL JOHNSON, PETITIONER v. UNITED STATES

on writ of certiorari to the united states court of appeals for the sixth circuit

[May 15, 2000]

Justice Souter delivered the opinion of the Court.

The issue in this case grows out of an Ex Post Facto Clause challenge to the retroactive application of 18 U. S. C. §3583(h), which authorizes a district court to impose an additional term of supervised release following the reimprisonment of those who violate the conditions of an initial term. The United States argues that district courts had the power to do so under the prior law, and that this cures any ex post facto problems. We agree with the Government as to the interpretation of prior law, and we find that consideration of the Ex Post Facto Clause is unnecessary.

I

In the Sentencing Reform Act of 1984, §212(a)(2), 98 Stat. 1999, Congress eliminated most forms of parole in favor of supervised release, a form of postconfinement monitoring overseen by the sentencing court, rather than the Parole Commission. See Gozlon-Peretz v. United States, 498 U. S. 395, 400–401 (1991). The sentencing court was authorized to impose a term of supervised release to follow imprisonment, with the maximum length of the termvarying according to the severity of the initial offense. See 18 U. S. C. §§3583(a), (b). While on supervised release, the offender was required to abide by certain conditions, some specified by statute and some imposable at the court’s discretion. See 18 U. S. C. §3583(d). Upon violation of a condition, 18 U. S. C. §3583(e)(3) (1988 ed., Supp. V) authorized the court to “revoke a term of supervised release, and require the person to serve in prison all or part of the term of supervised release without credit for time previously served on postrelease supervision … .” Such was done here.

In October 1993, petitioner Cornell Johnson violated 18 U. S. C. §1029(b)(2), a Class D felony. In March 1994, the United States District Court for the Eastern District of Tennessee sentenced him to 25 months’ imprisonment, to be followed by three years of supervised release, the maximum term available under §3583(b) for a Class D felony. Johnson was released from prison on August 14, 1995, having received good-conduct credits, and began serving his 3-year term of supervised release. Some seven months into that term, he was arrested in Virginia and later convicted of four state forgery-related offenses. He was thus found to have violated one of the conditions of supervised release made mandatory by §3583(d), that he not commit another crime during his term of supervised release, and one imposed by the District Court, that he not leave the judicial district without permission.

The District Court revoked Johnson’s supervised release, imposed a prison term of 18 months, and ordered Johnson placed on supervised release for 12 months following imprisonment. App. 40–41. For this last order, theDistrict Court did not identify the source of its authority, though under Circuit law it might have relied on §3583(h), a subsection added to the statute in 1994, see Violent Crime and Law Enforcement Act of 1996, §110505(2)(B), 108 Stat. 2017. Subsection (h) explicitly gave district courts the power to impose another term of supervised release following imprisonment, a power not readily apparent from the text of §3583(e)(3) (set out infra, at 8–9).

Johnson appealed his sentence, arguing that §3583(e)(3) gave district courts no such power and that applying §3583(h) to him violated the Ex Post Facto Clause of the Constitution, Art. I, §9. The Sixth Circuit, joining the majority of the Federal Courts of Appeals, had earlier taken Johnson’s position as far as the interpretation of §3583(e)(3) was concerned, holding that it did not authorize a district court to impose a new term of supervised release following revocation and reimprisonment. See United States v. Truss, 4 F. 3d 437 (CA6 1993). It nonetheless affirmed the District Court, judgt. order reported at 181 F. 3d 105 (1999), reasoning that the application of §3583(h) was not retroactive at all, since revocation of supervised release was punishment for Johnson’s violation of the conditions of supervised release, which occurred after the 1994 amendments. With no retroactivity, there could be no Ex Post Facto Clause violation. See App. 49 (citing United States v. Abbington, 144 F. 3d 1003, 1005 (CA6), cert. denied, 525 U. S. 933 (1998)). Other Circuits had held to the contrary, that revocation and reimprisonment were punishment for the original offense. From that perspective, application of §3583(h) was retroactive and at odds with the Ex Post Facto Clause. We granted certiorari to resolve the conflicts, 528 U. S. 590 (1999), and now affirm.

II

The heart of the Ex Post Facto Clause, U. S. Const., Art. I, §9, bars application of a law “that changes the punishment, and inflicts a greater punishment, than the law annexed to the crime, when committed … .” Calder v. Bull, 3 Dall. 386, 390 (1798) (emphasis deleted). To prevail on this sort of ex post facto claim, Johnson must show both that the law he challenges operates retroactively (that it applies to conduct completed before its enactment) and that it raises the penalty from whatever the law provided when he acted. See California Dept. of Corrections v. Morales, 514 U. S. 499, 506–507, n. 3 (1995).

A

The Sixth Circuit, as mentioned earlier, disposed of the ex post facto challenge by applying its earlier cases holding the application of §3583(h) not retroactive at all: revocation of supervised release “imposes punishment for defendants’ new offenses for violating the conditions of their supervised release.” United States v. Page, 131 F. 3d 1173, 1176 (1997). On this theory, that is, if the violation of the conditions of supervised release occurred after the enactment of §3583(h), as Johnson’s did, the new law could be given effect without applying it to events before its enactment.

While this understanding of revocation of supervised release has some intuitive appeal, the Government disavows it, and wisely so in view of the serious constitutional questions that would be raised by construing revocation and reimprisonment as punishment for the violation of the conditions of supervised release. Although such violations often lead to reimprisonment, the violative conduct need not be criminal and need only be found by a judge under a preponderance of the evidence standard, not by a jury beyond a reasonable doubt. See 18 U. S. C. §3583(e)(3) (1988 ed., Supp. V). Where the acts of violation are criminal in their own right, they may be the basis for separate prosecution, which would raise an issue of double jeopardy if the revocation of supervised release were also punishment for the same offense. Treating postrevocation sanctions as part of the penalty for the initial offense, however (as most courts have done), avoids these difficulties. See, e.g., United States v. Wyatt, 102 F. 3d 241, 244–245 (CA7 1996) (rejecting double jeopardy challenge on ground that sanctions for violating the conditions of supervised release are part of the original sentence); United States v. Beals, 87 F. 3d 854, 859–860 (CA7 1996) (noting that punishment for noncriminal violations must be justified by reference to original crimes), overruled on other grounds, United States v. Withers, 128 F. 3d 1167 (1997); United States v. Meeks, 25 F. 3d 1117, 1123 (CA2 1994) (noting absence of constitutional procedural protections in revocation proceedings). Cf. Gagnon v. Scarpelli, 411 U. S. 778, 782 (1973) (“Probation revocation … is not a stage of a criminal prosecution”). For that matter, such treatment is all but entailed by our summary affirmance of Greenfield v. Scafati, 277 F. Supp. 644 (Mass. 1967) (three-judge court), summarily aff’d, 390 U. S. 713 (1968), in which a three-judge panel forbade on ex post facto grounds the application of a Massachusetts statute imposing sanctions for violation of parole to a prisoner originally sentenced before its enactment. We therefore attribute postrevocation penalties to the original conviction.

B

Since postrevocation penalties relate to the original offense, to sentence Johnson to a further term of supervised release under §3583(h) would be to apply this section retroactively (and to raise the remaining ex post facto question, whether that application makes him worse off). But before any such application (and constitutional test), there is a question that neither party addresses. The Ex Post Facto Clause raises to the constitutional level one of the most basic presumptions of our law: legislation, especially of the criminal sort, is not to be applied retroactively. See, e.g., Lynce v. Mathis, 519 U. S. 433, 439 (1997); Landgraf v. USI Film Products, 511 U. S. 244, 265 (1994). Quite independent of the question whether the Ex Post Facto Clause bars retroactive application of §3583(h), then, there is the question whether Congress intended such application. Absent a clear statement of that intent, we do not give retroactive effect to statutes burdening private interests. See id., at 270.

The Government offers nothing indicating congressional intent to apply §3583(h) retroactively. The legislative decision to alter the rule of law established by the majority interpretation of §3583(e)(3) (no authority for supervised release after revocation and reimprisonment) does not, by itself, tell us when or how that legislative decision was intended to take effect. See Rivers v. Roadway Express, Inc., 511 U. S. 298, 304–307 (1994). Neither is there any indication of retroactive purpose in the omission of an express effective date from the statute. The omission simply remits us to the general rule that when a statute has no effective date, “absent a clear direction by Congress to the contrary, [it] takes effect on the date of its enactment.” Gozlon-Peretz, 498 U. S., at 404.

Nor, finally, has Congress given us anything expressly identifying the relevant conduct in a way that would point to retroactive intent. It may well be that Congress, like the Sixth Circuit, believed that §3583(h) would naturally govern sentencing proceedings for violations of supervised release that took place after the statute’s enactment, simply because the violation was the occasion for imposing the sanctions. But Congress gave us no clear indication to this effect, and we have already rejected that theory; the relevant conduct is the initial offense. In sum, there being no contrary intent, our longstanding presumption directs that §3583(h) applies only to cases in which that initial offense occurred after the effective date of the amendment, September 13, 1994.

Given this conclusion, the case does not turn on whether Johnson is worse off under §3583(h) than he previ-ously was under §3583(e)(3), as subsection (h) does not apply, and the ex post facto question does not arise. The case turns, instead, simply on whether §3583(e)(3) per-mitted imposition of supervised release following arecommitment.

III

Section 3583(e), at the time of Johnson’s conviction, authorized a district court to

“(1) terminate a term of supervised release and discharge the person released at any time after the expiration of one year of supervised release, pursuant to the provisions of the Federal Rules of Criminal Procedure relating to the modification of probation, if it is satisfied that such action is warranted by the conduct of the person released and the interest of justice;

“(2) extend a term of supervised release if less than the maximum authorized term was previously imposed, and … modify, reduce, or enlarge the conditions of supervised release, at any time prior to the expiration or termination of the term of supervised release, pursuant to the provisions of the Federal Rules of Criminal Procedure relating to the modification of probation and the provisions applicable to the initial setting of the terms and conditions of post-release supervision;

“(3) revoke a term of supervised release, and require the person to serve in prison all or part of the term of supervised release without credit for the time previously served on postrelease supervision, if it finds by a preponderance of the evidence that the person violated a condition of supervised release, pursuant to the provisions of the Federal Rules of Criminal Procedure that are applicable to probation revocation and to the provisions of applicable policy statements issued by the Sentencing Commission … .”

The text of subsection (e)(3) does not speak directly to the question whether a district court revoking a term of supervised release in favor of reimprisonment may require service of a further term of supervised release following the further incarceration. And if we were to concentrate exclusively on the verb “revoke,” we would not detect any suggestion that the reincarceration might be followed by another term of supervised release, the conventional understanding of “revoke” being simply “to annul by recalling or taking back.” Webster’s Third New International Dictionary 1944 (1981). There are reasons, nonetheless, to think that the option of further supervised release was intended.

First, there are some textual reasons, starting with the preceding subsection (e)(1). This is an unequivocal provision for ending the term of supervised release without the possibility of its reimposition or continuation at a later time. Congress wrote that when a court finds that a defendant’s conduct and the interests of justice warrant it, the court may “terminate a term of supervised release and discharge the person released,” once at least a year of release time has been served. If application of subsection (3) had likewise been meant to conclude any possibility of supervised release later, it would have been natural for Congress to write in like terms. It could have provided that upon finding a defendant in violation of the release conditions the Court could “terminate a term of supervised release” and order the defendant incarcerated for a term as long as the original supervised release term. But that is not what Congress did. Instead of using “terminate” with the sense of finality just illustrated in subsection (1), Congress used the verb “revoke” and so at the least left the door open to a reading of subsection (3) that wouldnot preclude further supervised release after the initial revocation. In fact, the phrasing of subsection (3) did more than just leave the door open to the non-preclusive reading.

As it was written before the 1994 amendments, subsection (3) did not provide (as it now does) that the court could revoke the release term and require service of a prison term equal to the maximum authorized length of a term of supervised release. It provided, rather, that the court could “revoke a term of supervised release, and require the person to serve in prison all or part of the term of supervised release … .” So far as the text is concerned, it is not a “term of imprisonment” that is to be served, but all or part of “the term of supervised release.” But if “the term of supervised release” is being served, in whole or part, in prison, then something about the term of supervised release survives the preceding order of revocation. While this sounds very metaphysical, the metaphysics make one thing clear: unlike a “terminated” order of supervised release, one that is “revoked” continues to have some effect. And since it continues in some sense after revocation even when part of it is served in prison, why can the balance of it not remain effective as a term of supervised release when the reincarceration is over?

Without more, we would have to admit that Congress had used “revoke” in an unconventional way in subsection (3), but it turns out that the unconventional sense is not unheard of. See United States v. O’Neil, 11 F. 3d 292, 295–296 (CA1 1993). Webster’s Third New International Dictionary (our edition of which was issued three years before the 1984 Act) reveals that “revoke” can mean “to call or summon back,” without the implication (here) that no further supervised release is subsequently possible. It gives “recall” as a synonym and comments that “RECALL in this sense indicates a calling back, suspending, or abrogating, either finally as erroneous or ill-advised or tentatively for deliberation … .” Ibid. The unconventional dictionary definition is not, of course, dispositive (although the emphasis placed upon it by Justice Scalia might suggest otherwise, see post., at 4–5). What it does do, however, is to soften the strangeness of Congress’s unconventional sense of “revoke” as allowing a “revoked” term of supervised release to retain vitality after revocation. It shows that saying a “revoked” term of supervised release survives to be served in prison following the court’s reconsideration of it is consistent with a secondary but recognized definition, and so is saying that any balance not served in prison may survive to be served out as supervised release.

A final textually based point is that the result of recognizing Congress’s unconventional usage of “revoke” is far less remarkable even than the unconventional usage. Let us suppose that Congress had legislated in language that unequivocally supported the dissent, by writing subsection (3) to provide that the judge could “revoke” or “terminate” the term of supervised release and sentence the defendant to a further term of incarceration. There is no reason to think that under that regime the court would lack the power to impose a subsequent term of supervised release in accordance with its general sentencing authority under 18 U. S. C. §3583(a). This section provides that “[t]he court, in imposing a sentence to a term of imprisonment for a felony or a misdemeanor, may include as a part of the sentence a requirement that the defendant be placed on a term of supervised release after imprisonment … .” Thus, on the dissent’s reading, when Johnson’s supervised release was revoked and he was committed to prison, the District Court “impos[ed] a sentence to a term of imprisonment.” See, e.g., App. 36, 39. And that sentence was, as already noted, imposed for his initial offense, the Class D felony violation of §1029(b)(2). See supra, at 4–6. Nor would it be mere formalism to link the second prison sentence to the initial offense; the gravity of the initial offense determines the maximum term of reimprisonment, see §3583(e)(3), just as it controls the maximum term of supervised release in the initial sentencing, see §3583(b). Since on the dissent’s understanding the resentencing proceeding would fall literally and sensibly within the terms of §3583(a), a plain meaning approach would find authority for reimposition of supervised release there. Cf. United States v. Wesley, 81 F. 3d 482, 483–484 (CA4 1996) (finding that §3583(a) grants power to impose a term of supervised release following reimprisonment at resentencing for violation of probation).

There is, then, nothing surprising about the consequences of our reading. The reading also enjoys the virtue of serving the evident congressional purpose. The congressional policy in providing for a term of supervised release after incarceration is to improve the odds of a successful transition from the prison to liberty. See, e.g., United States v. Johnson, 529 U. S. ___, ___ (2000) (slip op., at 6) (“Congress intended supervised release to assist individuals in their transition to community life. Supervised release fulfills rehabilitative ends, distinct from those served by incarceration.”). The Senate Report was quite explicit about this, stating that the goal of supervised release is “to ease the defendant’s transition into the community after the service of a long prison term for a particularly serious offense, or to provide rehabilitation to a defendant who has spent a fairly short period in prison for punishment or other purposes but still needs supervision and training programs after release.” S. Rep. No. 98–225, p. 124 (1983).

Prisoners may, of course, vary in the degree of help needed for successful reintegration. Supervised release departed from the parole system it replaced by giving district courts the freedom to provide post-release supervision for those, and only those, who needed it. See id., at 125 (“In effect, the term of supervised release provided by the bill takes the place of parole supervision under current law. Unlike current law, however, probation officers will only be supervising those releasees from prison who actually need supervision, and every releasee who does need supervision will receive it”). Congress aimed, then, to use the district courts’ discretionary judgment to allocate supervision to those releasees who needed it most. But forbidding the reimposition of supervised release after revocation and reimprisonment would be fundamentally contrary to that scheme. A violation of the terms of supervised release tends to confirm the judgment that help was necessary, and if any prisoner might profit from the decompression stage of supervised release, no prisoner needs it more than one who has already tried liberty and failed. He is the problem case among problem cases, and a Congress asserting that “every releasee who does need supervision will receive it,” ibid., seems very unlikely to have meant to compel the courts to wash their hands of the worst cases at the end of reimprisonment.

The idea that a sentencing court should have authority to subject a reincarcerated prisoner to further supervised release has support, moreover, in the pre-Guidelines practice with respect to nondetentive monitoring, as illuminated in United States v. O’Neil, 11 F. 3d 292 (CA1 1993). The Sentencing Guidelines, after all, “represent an approach that begins with, and builds upon,” pre-Guidelines law, see USSG, ch. 1, pt. A, intro. comment. 3, and when a new legal regime develops out of an iden-tifiable predecessor, it is reasonable to look to the pre-cursor in fathoming the new law. Cf. INS v. Cardoza&nbhyph;Fonseca, 480 U. S. 421, 432–434 (1987) (examining practice under precursor statute to determine meaning of amended statute).

Two sorts of nondetentive monitoring existed before the introduction of supervised release: probation and parole. Of these pre-Guidelines options, the one more closely analogous to supervised release following imprisonment was parole, which by definition was a release under supervision of a parole officer following service of some term of incarceration. Courts have commented on the similarity. See, e.g., Meeks, 25 F. 3d, at 1121 (“[S]upervised release is essentially similar to parole”); United States v. Paskow, 11 F. 3d 873, 881 (CA9 1993) (“Supervised release and parole are virtually identical systems”).

In thinking about this case, it is striking that the provisions of the former parole scheme dealing with the consequences of violating parole conditions repeatedly used the verb “revoke.” See, e.g., 18 U. S. C. §4214(d)(5) (1982 ed.) (repealed 1984, Pub. L. 98–473, §§218(a)(5), 235, 98 Stat. 2027, 2031) (revocation of parole); 21 U. S. C. §841(c) (1982 ed.) (repealed 1984) (revocation of special parole). And yet there seems never to have been a question that a new term of parole could follow a prison sentence imposed after revocation of an initial parole term. See, e.g., 28 CFR §2.52(b) (1999) (following revocation of parole, Sentencing Commission will determine whether reparole is warranted); O’Neil, supra, at 299; United States Parole Comm’n v. Williams, 54 F. 3d 820, 824 (CADC 1995) (noting “the established pre-Guidelines sentencing principle that parole is available unless expressly precluded” (citation and internal quotation marks omitted)). Thus, “revocation” of parole followed by further imprisonment was not a mere termination of a limited liberty that a defendant could experience only once per conviction, and it is fair to suppose that in the absence of any textual bar “revocation” of parole’s replacement, supervised release, was meant to leave open the possibility of further supervised release, as well.

As seen already, “revoke” is no such bar, and we find no other. The proceeding that follows a violation of the conditions of supervised release is not, to be sure, a precise reenactment of the initial sentencing. Section 3583(e)(3) limits the possible prison term to the duration of the term of supervised release originally imposed. (If less than the maximum has been imposed, a court presumably may, before revoking the term, extend it pursuant to §3583(e)(2); this would allow the term of imprisonment to equal the term of supervised release authorized for the initial offense.) The new prison term is limited further according to the gravity of the original offense. See §3583(e)(3). But nothing in these specific provisions suggests that the possibility of supervised release following imprisonment was meant to be eliminated.

In sum, from a purely textual perspective, the more plausible reading of §3583(e)(3) before its amendment and the addition of subsection (h) leaves open the possibility of supervised release after reincarceration. Pre-Guidelines practice, linguistic continuity from the old scheme to the current one, and the obvious thrust of congressional sentencing policy confirm that, in applying the law as before the enactment of subsection (h), district courts have the authority to order terms of supervised release following reimprisonment.

The judgment of the Court of Appeals for the Sixth Circuit is

Affirmed.

 The current version of §3583(e)(3) reads slightly differently, but for reasons discussed below, we focus on the law in effect at the time of Johnson’s initial crime.

 Of the 11 Circuits to consider the issue, 9 had reached this conclusion. See, e.g., United States v. Koehler, 973 F. 2d 132 (CA2 1992); United States v. Malesic, 18 F. 3d 205 (CA3 1994); United States v. Cooper, 962 F. 2d 339 (CA4 1992); United States v. Holmes, 954 F. 2d 270 (CA5 1992); United States v. Truss, 4 F. 3d 437 (CA6 1993); United States v. McGee, 981 F. 2d 271 (CA7 1992); United States v. Behnezhad, 907 F. 2d 896 (CA9 1990); United States v. Rockwell, 984 F. 2d 1112 (CA10 1993); United States v. Tatum, 998 F. 2d 893 (CA11 1993). Two, the First and the Eighth, found that §3583(e)(3) did grant district courts such power. See United States v. O’Neil, 11 F. 3d 292 (CA1 1993); United States v. Schrader, 973 F. 2d 623 (CA8 1992).

 See, e.g., United States v. Eske, 189 F. 3d 536, 539 (CA7 1999); United States v. Lominac, 144 F. 3d 308, 312 (CA4 1998); United States v. Dozier, 119 F. 3d 239, 241 (CA3 1997); United States v. Collins, 118 F. 3d 1394, 1397 (CA9 1997); United States v. Meeks, 25 F. 3d 1117, 1124 (CA2 1994) (addressing §3583(g)). In contrast to these cases, the First and Eighth Circuits, relying on their broader construction of §3583(e)(3), concluded that application of §3583(h) did not violate the Ex Post Facto Clause. See United States v. Sandoval, 69 F. 3d 531 (CA1 1995) (unpublished), cert. denied, 519 U. S. 821 (1996); United States v. St. John, 92 F. 3d 761 (CA8 1996).

 Indeed, the Sentencing Guidelines identify the effective date of §3583(h) as September 13, 1994. United States Sentencing Commission, Guidelines Manual §7B1.3, comment., n. 2 (Nov. 1998) (USSG). So, too, have the federal courts. See, e.g., United States v. Hale, 107 F. 3d 526, 529, n. 3 (CA7 1997).

 The failure to specify an effective date evidences at least arguable diffidence on this point. Another section of the same Act that added §3583(h) amended 18 U. S. C. §3553 to limit the applicability of some statutory minimum sentences. See §80001, 108 Stat. 1985. That amendment, the section made explicit, “shall apply to all sentences imposed on or after the 10th day beginning after the date of enactment of this Act.” §80001(c), 108 Stat. 1986.

 We took a similar approach in Cisneros v. Alpine Ridge Group, 508 U. S. 10 (1993). The respondents in that case were private developers who had entered into contracts with the Department of Housing and Urban Development. When the Department sought to recalibrate payments it owed under the contracts, the developers sued, and the Ninth Circuit ruled that the Department’s proposed method of calculating payments was prohibited by the contracts. Congress subsequently passed legislation explicitly authorizing that method of calculation. The developers resisted application of that legislation to their contracts on the grounds that it retroactively deprived them of vested contractual rights, in violation of the Due Process Clause. We ruled (disagreeing with the Ninth Circuit’s earlier holding) that the Department’s methodology was acceptable under the contracts as signed. Finding the governmental action permitted by the old law, we declined to consider the constitutional consequences of a legislative attempt to change the applicable law.

 The dissent offers an erudite explanation of the different senses of the two words, intending to demonstrate that Congress displayed “an admirably precise use of language,” by using “revoke” to mean “annul” and “terminate” to indicate that “[t]he supervised release is treated as fulfilled, and the sentence is complete.” Post, at 3–4 (opinion of Scalia, J.). That is virtuoso lexicography, but it shows only that English is rich enough to give even textualists room for creative readings. This one encounters serious difficulties; the very same section of the statute (as in effect at the time of Johnson’s offense) provides that if the person released is found in possession of a controlled substance, “the court shall terminate the term of supervised release and require the defendant to serve in prison not less than one-third of the term of supervised release.” 18 U. S. C. §3583(g) (1988 ed.).

 Justice Scalia, post at 7, thinks the “term” survives only as a measure of duration, but of course the statute does not read “require the person to serve a term in prison equal to all or part of the term of supervised release … .”

 While this sense is of course less common, the most recent editions of the most authoritative dictionaries do not tag it as rare or obsolete. The Oxford English Dictionary gives five examples of this usage, albeit hardly recent ones: three are drawn from the late 16th century and the most recent from 1784. 13 Oxford English Dictionary 838 (2d ed. 1989). But the OED is unabashedly antiquarian; of its examples for the more common meaning of “revoke,” the most recent dates from 1873. Ibid. Webster’s, it should be noted, includes the less common meaning, without antiquarian reproach, in its third edition. Webster’s Third New International Dictionary 1944 (1981).

As Justice Scalia remarks, in relying on an uncommon sense of the word, we are departing from the rule of construction that prefers ordinary meaning, see post, at 1. But this is exactly what ought to happen when the ordinary meaning fails to fit the text and when the realization of clear congressional policy (here, favoring the ability to impose supervised release) is in tension with the result that customary interpretive rules would deliver. See, e.g., Commissioner v. Brown, 380 U. S. 563, 571 (1965) (recognizing “some ‘scope for adopting a restricted rather than a literal or usual meaning of its words where acceptance of that meaning … would thwart the obvious purpose of the statute’ ”) (quoting Helvering v. Hammel, 311 U. S. 504, 510–511 (1941); In re Chapman, 166 U. S. 661, 667 (1897) (“[N]othing is better settled, than that statutes should receive a sensible construction, such as will effectuate the legislative intention, and, if possible, so as to avoid an unjust or an absurd conclusion”). When text implies that a word is used in a secondary sense and clear legislative purpose is at stake, Justice Scalia’s cocktail-party textualism, post, at 4, must yield to the Congress of the United States. (Not that we consider usage at a cocktail party a very sound general criterion of statutory meaning: a few nips from the flask might actually explain the solecism of the dissent’s gunner who “revoked” his bird dog, post, at 6, n. 4; in sober moments he would know that dogs cannot be revoked, even though sentencing orders can be. His mistake, in any case, tells us nothing about how Congress may have used “revoke” in the statute. The gunner’s error is, as Justice Scalia notes, one of current usage. (It is not merely that we do not “revoke” dogs in a “literal” sense today, as Justice Scalia puts it; we do not revoke them at all.) The question before us, however, is one of definition as distinct from usage: when Congress employed the modern usage in providing that a term of supervised release could be revoked, was it employing the most modern meaning of the term “revoke”? Usage can be a guide but not a master in answering a question of meaning like this one. Justice Scalia’s argument from the current unacceptability of the dog and ox examples thus jeopardizes sound statutory construction rather more severely than his sportsman ever threatened a bird.)

 Justice Scalia attributes the strong preference for supervised release at the conclusion of a prison term to this Court, post, at 10, when that view of penal policy comes not from the Court but from Congress. The point is crucial. Our obligation is to give effect to congressional purpose so long as the congressional language does not itself bar that result. See, e.g., Holloway v. United States, 526 U. S. 1, 9 (1999) (noting that statutory language should be interpreted in light of congressional policy); Caron v. United States, 524 U. S. 308, 315 (1998) (rejecting petitioner’s reading of a statute because it “yields results contrary to a likely, and rational, congressional policy”). One who believes that courts must not look beyond text might well find any invocation of policy unjustified (even willful), at least when the policy does not rise unbidden from the words of the statute, but we have never treated the text as such a jealous guide and have traditionally sought to construe a statute so as to reach results consistent with what Chief Justice Taney called “its object and policy.” See United States v. Heirs of Boisdoré, 8 How. 113, 122 (1849). And in what Chief Justice Marshall called the attempt “to discover the design of the legislature,” we have “seize[d] every thing from which aid can be derived.” United States v. Fisher, 2 Cranch 358, 386 (1805).

 The same is true of special parole, part of the required sentence for certain drug offenses. Though the special parole statute did not explicitly authorize reimposition of special parole after revocation of the initial term and reimprisonment, the Parole Commission required it. See 28 CFR §2.57(c) (1999). Some courts have recently decided that this regulation is inconsistent with 21 U. S. C. §841(c) (1982 ed.), see, e.g., Evans v. United States Parole Comm’n, 78 F. 3d 262 (CA7 1996), but this does not affect the backdrop against which Congress legislated in 1984.

As for probation, the sentencing court’s power to order a new term following revocation was the subject of some disagreement. The pre-Guidelines statute authorized the court to “revoke the probation and … impose any sentence which might originally have been imposed.” 18 U. S. C. §3653 (1982 ed.) (repealed). The statute thus clearly specified that the options for post-revocation sentencing were those available at the original sentencing; courts disputed only whether probation was a “sentence” that could be imposed. See O’Neil, 11 F. 3d, at 298–299 (collecting cases). The dispute over what counted as a sentence does not affect the broader point that a court’s powers at the original sentencing are the baseline from which powers at resentencing are determined. Nor is our analysis of supervised release drawn into question by the fact that courts could not, for violations of probation, impose imprisonment followed by probation. Probation, unlike supervised release, was an alternative to imprisonment. Courts did not have the power to impose both at the original sentencing, so their inability to do so at subsequent sentencings is no surprise.

 The dissent seems to misconstrue our discussion of pre-Guidelines practice, see post, at 11–12, claiming that the practice is unilluminating because the possibility of parole inhered in any prison sentence. But our point simply is that, metaphysics aside, Congress gave no indication that it thought supervised release after reincarceration would be less valuable than reparole after reincarceration had been.

 Nor does our traditional rule of lenity in interpreting criminal statutes demand a contrary result. Lenity applies only when the equipoise of competing reasons cannot otherwise be resolved (not the case here), and in any event the rule of lenity would be Delphic in this case. There is simply no way to tell whether sentencing courts given the option of supervised release will generally be more or less lenient in fixing the second prison sentence.