delivered the opinion of the Court.
Under the Armed Career Criminal Act (ACCA), 18 U. S. C. § 924(e)(2)(A)(ii), a state drug-trafficking conviction qualifies as “a serious drug offense” if “a maximum term of imprisonment of ten years or more is prescribed by law” for the “offense.” The Court of Appeals for the Ninth Circuit held that “the maximum term of imprisonment . . . prescribed by law” must be determined without taking recidivist enhancements into account. 464 F. 3d 1072,1082 (2006). We reverse.
I
At issue in this case is respondent’s sentence on his 2004 conviction in the United States District Court for the Eastern District of Washington for possession of a firearm by a convicted felon, in violation of 18 U. S. C. § 922(g)(1). Respondent had two prior state convictions in California for residential burglary and three state convictions in Washington for delivery of a controlled substance, in violation of Wash. Rev. Code §§69.50.401(a)(l)(ii)-(iv) (1994). 1 Respondent’s three Washington drug convictions occurred on the same day but were based on deliveries that took place on three separate dates. Sentencing Order in No. CR-03-142*381RHW (ED Wash., Sept. 3, 2004), p. 5, App. 245, 250 (hereinafter Sentencing Order). At the time of respondent’s drug offenses, the Washington statute that respondent was convicted of violating stated that, upon conviction, a defendant could be “imprisoned for not more than five years,” §§ 69.50.401(a)(l)(ii)-(iv), but another provision specified that “[a]ny person convicted of a second or subsequent offense” could “be imprisoned for a term up to twice the term otherwise authorized,” § 69.50.408(a). Thus, by virtue of this latter, recidivist, provision respondent faced a maximum penalty of imprisonment for 10 years. The judgment of conviction for each of the drug-delivery charges listed the maximum term of imprisonment for the offense as “ten years,” App. 16, 42, 93, but the state court sentenced respondent to concurrent sentences of 48 months’ imprisonment on each count, id., at 21, 47, 98.
In the federal felon-in-possession case, the Government asked the District Court to sentence respondent under ACC A, which sets a 15-year minimum sentence “[i]n the case of a person who violates section 922(g) of [Title 18] and has three previous convictions ... for a violent felony or a serious drug offense, or both, committed on occasions different from one another . . . .” 18 U. S. C. § 924(e)(1) (2000 ed., Supp. V). The Government argued that respondent’s two prior California burglary convictions were for “ Violent felonies.’” Pet. for Cert. 4. See § 924(e)(2)(B)(ii) (2000 ed.) (listing “burglary” as a “violent felony”). The District Court agreed, and that ruling is not at issue here.
The Government also argued that at least two of respondent’s Washington drug convictions were for “serious drug offense[s].” Under ACC A, a “serious drug offense” includes:
“an offense under State law, involving manufacturing, distributing, or possessing with intent to distribute, a controlled substance (as defined in section 102 of the Controlled Substances Act (21 U. S. C. 802)), for which *382a maximum term of imprisonment of ten years or more is prescribed by law.” § 924(e)(2)(A)(ii) (emphasis added).
Because the maximum term that respondent faced on at least two of the Washington charges was 10 years, the Government contended that these convictions had to be counted under ACCA. The District Court disagreed, holding that respondent’s drug-trafficking convictions were not convictions for “serious drug offense[s]” under ACCA because the “maximum term of imprisonment” for the purposes of § 924(e)(2)(A)(ii) is determined without reference to recidivist enhancements. Sentencing Order, at 9, App. 254.
The Court of Appeals for the Ninth Circuit, applying its prior precedent in United States v. Corona-Sanchez, 291 F. 3d 1201 (2002) (en banc), affirmed. 464 F. 3d 1072. The court recognized that its decision conflicted with the Seventh Circuit’s decision in United States v. Henton, 374 F. 3d 467, 469-470, cert. denied, 543 U. S. 967 (2004), and was “in tension” with decisions of the Fourth and Fifth Circuits. 464 F. 3d, at 1082, n. 6; see Mutascu v. Gonzales, 444 F. 3d 710, 712 (CA5 2006) (per curiam); United States v. Williams, 326 F. 3d 535, 539 (CA4 2003). We granted the Government’s petition for a writ of certiorari, 551 U. S. 1191 (2007).
II
The question that we must decide is whether the “maximum term of imprisonment prescribed by law” in this case is, as respondent maintains and the Ninth Circuit held, the 5-year ceiling for first offenses or, as the Government contends, the 10-year ceiling for second or subsequent offenses. See Wash. Rev. Code §§ 69.50.401(a)(1)(ii)-(iv), 69.50.408(a).
The Government’s reading is compelled by the language of ACCA. For present purposes, there are three key statutory terms: “offense,” “law,” and “maximum term.” The “offense” in each of the drug-delivery cases was a violation of §§ 69.50.401(a)(l)(ii)-(iv). The relevant “law” is set out in *383both that provision, which prescribes a “maximum term” of 5 years for a first “offense,” and § 69.50.408(a), which prescribes a “maximum term” of 10 years for a second or subsequent “offense.” Thus, in this case, the maximum term prescribed by Washington law for at least two of respondent’s state drug offenses was 10 years.
The Ninth Circuit’s holding that the maximum term was five years contorts ACCA’s plain terms. Although the Washington state court sentenced respondent to 48 months’ imprisonment, there is no dispute that § 69.50.408(a) permitted a sentence of up to 10 years. On the Ninth Circuit’s reading of ACCA, even if respondent had been sentenced to, say, six years’ imprisonment, “the maximum term of imprisonment” prescribed by law still would have been five years. It is hard to accept the proposition that a defendant may lawfully be sentenced to a term of imprisonment that exceeds the “maximum term of imprisonment . . . prescribed by law,” but that is where the Ninth Circuit’s reading of the statute leads.
The Ninth Circuit’s interpretation is also inconsistent with the way in which the concept of the “maximum term of imprisonment” is customarily understood by participants in the criminal justice process. Suppose that a defendant who indisputably had more than three prior convictions for “violent felon[ies]” or “serious drug offense[s]” was charged in federal court with violating the felon-in-possession statute. Under ACCA, this defendant would face a sentence of “not less than 15 years.” 18 U. S. C. § 924(e)(1) (2000 ed., Supp. V). Suppose that the defendant asked his or her attorney, “What’s the maximum term I face for the new offense?” An attorney aware of ACCA would surely not respond, “10 years,” even though 10 years is the maximum sentence without the ACCA enhancement. See § 924(a)(2) (2000 ed.).
Suppose that the defendant then pleaded guilty to the felon-in-possession charge. Under Federal Rule of Criminal Procedure 11(b)(1)(H), the trial judge would be required to *384advise the defendant of the “maximum possible penalty.” If the judge told the defendant that the maximum possible sentence was 10 years and then imposed a sentence of 15 years based on ACCA, the defendant would have been sorely misled and would have a ground for moving to withdraw the plea. See United States v. Gonzalez, 420 F. 3d 111, 132 (CA2 2005); United States v. Harrington, 354 F. 3d 178, 185-186 (CA2 2004). In sum, a straightforward application of the language of ACCA leads to the conclusion that the “maximum term of imprisonment prescribed by law” in this case was 10 years.
Ill
A
In an effort to defend the Ninth Circuit’s decision, respondent offers both a textual argument and a related argument based on the “manifest purpose” of ACCA. Brief for Respondent 8.
Respondent’s textual argument is as follows. The term “offense” “generally is understood to describe the elements constituting the crime.” Id., at 10. Because prior convictions required for recidivist enhancements are not typically offense elements, they should not be considered part of the “offense” under ACCA. Thus, the “maximum term of imprisonment prescribed by law” for the drug convictions at issue was the maximum term prescribed for simply committing the elements of the drug offense and was therefore five years. Id., at 10-11.
Respondent’s argument is not faithful to the statutory text. Respondent reads ACCA as referring to “the maximum term of imprisonment prescribed by law” for a defendant with no prior convictions that trigger a recidivist enhancement, but that is not what ACCA says. ACCA instead refers to “the maximum term of imprisonment prescribed by law” for “an offense,” and, as previously explained, in this case, the maximum term prescribed by Washington law for each of respondent’s two relevant offenses was 10 years.
*385Respondent’s argument based on ACCA’s “manifest purpose” must also be rejected. Respondent argues that ACCA uses “the maximum penalty specified for the offense by state law as a short-hand means of identifying conduct deemed sufficiently ‘serious’ to trigger [the] mandatory penalty.” Id., at 9. According to respondent, “[t]he nature of [a defendant’s] conduct, the elements of the offense, and the impact of the crime ... are the characteristics that typically are used to gauge the ‘seriousness’ of an offense,” and a defendant’s “status as a recidivist has no connection to whether the offense committed by the defendant was a ‘serious’ one.” Id., at 11.
This argument rests on the erroneous proposition that a defendant’s prior record of convictions has no bearing on the seriousness of an offense. On the contrary, however, an offense committed by a repeat offender is often thought to reflect greater culpability and thus to merit greater punishment. Similarly, a second or subsequent offense is often regarded as more serious because it portends greater future danger and therefore warrants an increased sentence for purposes of deterrence and incapacitation. See Witte v. United States, 515 U. S. 389, 403 (1995); Spencer v. Texas, 385 U. S. 554, 570 (1967) (Warren, C. J., dissenting in two judgments and concurring in one).
If respondent were correct that a defendant’s record of prior convictions has no bearing on the seriousness of an offense, then it would follow that any increased punishment imposed under a recidivist provision would not be based on the offense of conviction but on something else — presumably the defendant’s prior crimes or the defendant’s “status as a recidivist,” Brief for Respondent 11. But we have squarely rejected this understanding of recidivism statutes. In Nichols v. United States, 511 U. S. 738 (1994), we explained that “ ‘[t]his Court consistently has sustained repeat-offender laws as penalizing only the last offense committed by the defendant.’” Id., at 747 (quoting Baldasar v. Illinois, 446 *386U. S. 222, 232 (1980) (Powell, J., dissenting)). When a defendant is given a higher sentence under a recidivism statute — or for that matter, when a sentencing judge, under a guidelines regime or a discretionary sentencing system, increases a sentence based on the defendant’s criminal history — 100% of the punishment is for the offense of conviction. None is for the prior convictions or the defendant’s “status as a recidivist.” The sentence “is a stiffened penalty for the latest crime, which is considered to be an aggravated offense because [it is] a repetitive one.” Gryger v. Burke, 334 U. S. 728, 732 (1948).
B
Respondent argues that our interpretation of ACCA produces “a sort of perverse bootstrapping” under which a defendant is “punished under federal law for being treated as a recidivist under state law,” Brief for Respondent 20 (emphasis deleted), but the fact that ACCA is itself a recidivist statute bolsters our reading. Since ACCA is a recidivist statute, Congress must have had such provisions in mind and must have understood that the “maximum penalty prescribed by [state] law” in some cases would be increased by state recidivism provisions.
Contrary to respondent’s suggestion, United States v. LaBonte, 520 U. S. 751 (1997), supports our interpretation of ACCA. The statute at issue in LaBonte, a provision of the Sentencing Reform Act of 1984, as amended, 28 U. S. C. § 994(h), directed the United States Sentencing Commission to “assure” that the Sentencing Guidelines specify a prison sentence “at or near the maximum term authorized for categories of” adult offenders who commit their third felony drug offense or violent crime. We held that the phrase “maximum term authorized” “refers to all applicable statutes,” including recidivist enhancements. 520 U. S., at 758, n. 4.
Respondent claims that LaBonte supports his position because ACCA, unlike 28 U. S. C. § 994(h), does not refer to *387“categories of” offenders. Respondent suggests that Congress’ failure to include such language in ACCA means that Congress intended to refer to a “maximum term” that does not depend on whether a defendant falls into the first-time-offender or recidivist “category.” Respondent does not explain how 18 U. S. C. § 924(e)(2)(A) could have easily been reworded to mirror 28 U. S. C. § 994(h). But in any event, the language used in ACCA, for the reasons explained above, is more than clear enough.
Respondent argues that the Ninth Circuit’s decision is supported by the so-called “categorical” approach that we used in Taylor v. United States, 495 U. S. 575 (1990), in determining which offenses qualify as “violent felon[ies]” under 18 U. S. C. § 924(e)(2)(B)(ii). Section 924(e)(2)(B)(ii) provides that four enumerated crimes — burglary, arson, extortion, and offenses involving the use of explosives — are “violent felonies]” for ACCA purposes. In Taylor, we held that Congress intended for these crimes to have a “uniform definition” that was “independent of the labels employed by the various States’ criminal codes.” 495 U. S., at 592. According to respondent, “[t]he categorical approach rests on the congressional intent — reflected in the statutory language— to focus the ACCA inquiry on the offense of conviction, rather than on collateral matters unrelated to the definition of the crime.” Brief for Respondent 12.
We see no connection, however, between the issue in Taylor (the meaning of the term “burglary” in § 924(e)(2)(B)(ii)) and the issue here (the meaning of the phrase “maximum term of imprisonment . . . prescribed by law” under § 924(e)(2)(A)(ii)). Taylor held that the meaning of “burglary” for purposes of ACCA does not depend on the label attached by the law of a particular State, 495 U. S., at 600-601, but the “maximum penalty prescribed by law” for a state offense necessarily depends on state law.
For a similar reason, we reject respondent’s argument that, under our interpretation, offenses that are not really *388serious will be included as “serious drug offense[s]” because of recidivist enhancements. In § 924(e)(2)(A)(ii), Congress chose to rely on the “maximum term of imprisonment . . . prescribed” by state law as the measure of the seriousness of state offenses involving the manufacture, distribution, or possession of illegal drugs. Congress presumably thought — not without reason — that if state lawmakers provide that a crime is punishable by 10 years’ imprisonment, the lawmakers must regard the crime as “serious,” and Congress chose to defer to the state lawmakers’ judgment. Therefore, our interpretation poses no risk that a drug-trafficking offense will be treated as “serious” without satisfying the standard that Congress prescribed.2
C
Respondent argues that it will often be difficult to determine whether a defendant faced the possibility of a recidivist enhancement in connection with a past state drug conviction and that therefore our interpretation of ACCA will require the federal courts to “engage in difficult inquiries regarding novel questions of state law and complex factual determinations about long-past proceedings in state courts.” Brief for Respondent 21. Respondent greatly exaggerates the problems to which he refers.
*389First, in some cases, a defendant will have received a recidivist enhancement, and this will necessarily be evident from the length of the sentence imposed. Second, as the present case illustrates, see App. 16, 42, 93, the judgment of conviction will sometimes list the maximum possible sentence even where the sentence that was imposed did not exceed the top sentence allowed without any recidivist enhancement. Third, as respondent himself notes, some jurisdictions require that the prosecution submit a formal charging document in order to obtain a recidivist enhancement. See Brief for Respondent 33. Such documents fall within the limited list of generally available documents that courts already consult for the purpose of determining if a past conviction qualifies as an ACCA predicate. See Shepard v. United States, 544 U. S. 13, 20 (2005). Fourth, in those cases in which the defendant pleaded guilty to the state drug charges, the plea colloquy will very often include a statement by the trial judge regarding the maximum penalty. This is mandated by Federal Rule of Criminal Procedure 11(b)(1)(H), and many States have similar requirements.3 Finally, in those cases in which the records that may properly be consulted do not show that the defendant faced the possibility of a recidivist enhancement, it may well be that the Government will be precluded from establishing that a conviction was for a qualifying offense. The mere possibility that some future cases might present difficulties cannot justify a reading of ACCA that disregards the clear meaning of the statutory language.
*390D
Respondent’s last argument is that if recidivist enhancements can increase the “maximum term” of imprisonment under ACCA, it must follow that mandatory guidelines systems that cap sentences can decrease the “maximum term” of imprisonment. Brief for Respondent 38. In each situation, respondent argues, the “maximum term” of imprisonment is the term to which the state court could actually have sentenced the defendant. Respondent concedes that he has waived this argument with respect to his own specific state-court convictions. See Brief in Opposition 15, n. 7. He argues, however, that Congress cannot have wanted to make the “maximum term” of imprisonment for ACCA purposes dependent on the complexities of state sentencing guidelines. We conclude, however, that the phrase “maximum term of imprisonment . . . prescribed by law” for the “offense” was not meant to apply to the top sentence in a guidelines range.
First, the top sentence in a guidelines range is generally not really the “maximum term . . . prescribed by law” for the “offense” because guidelines systems typically allow a sentencing judge to impose a sentence that exceeds the top of the guidelines range under appropriate circumstances. The United States Sentencing Guidelines, for example, permit “upward departures,” see United States Sentencing Commission, Guidelines Manual § 5K2.0 (Nov. 2007), and essentially the same characteristic was shared by all of the mandatory guidelines systems in existence at the time of the enactment of the ACCA provision at issue in this case.4 *391(Following this pattern, Washington law likewise provided at the time of respondent’s state convictions that a sentencing judge could “impose a sentence outside the standard sentence range” upon a finding “that there [were] substantial and compelling reasons justifying an exceptional sentence.” Wash. Rev. Code § 9.94A.120(2) (1994).5)
Second, the concept of the “maximum” term of imprisonment or sentence prescribed by law was used in many statutes that predated the enactment of ACCA and the federal Sentencing Reform Act of 1984, Pub. L. 98-473, §211, 98 Stat. 1987, and in all those statutes the concept necessarily referred to the maximum term prescribed by the relevant criminal statute, not the top of a sentencing guidelines range. See, e. g., 18 U. S. C. § 3 (1982 ed.) (“[A]n accessory after the fact shall be imprisoned not more than one-half the maximum term of imprisonment ... for the punishment of the principal”); § 3575(b) (allowing for an increased sentence for dangerous special offenders “not disproportionate in severity to the maximum term otherwise authorized by law for” the underlying felony); see also §371 (the punishment for conspiracy to commit a misdemeanor “shall not exceed the maximum punishment provided for such misdemeanor”); §3651 (allowing for confinement and suspension of sentence upon conviction of an offense not punishable by death or life imprisonment “if the maximum sentence for such offense is *392more than six months”); § 3653 (referring to the “maximum probation period”).
It is instructive that, even in the Sentencing Reform Act, the concept of the “maximum term of imprisonment” prescribed for an offense was used in this sense. See § 212, 98 Stat. 1991-1992 (new 18 U. S. C. § 3559 classifying offenses based on “the maximum term of imprisonment authorized... by the statute describing the offense”); § 235(b)(1)(F), 98 Stat. 2032 (“The maximum term of imprisonment in effect on the effective date [of the Sentencing Reform Act]” remains in effect for five years after the effective date “for an offense committed before the effective date”); § 1003(a), id., at 2138 (solicitation to commit a crime of violence punishable by “one-half the maximum term of imprisonment... prescribed for the punishment of the crime solicited”). In light of this established pattern and the relative newness of sentencing guidelines systems when the ACCA provision at issue here was added, we conclude that Congress meant for the concept of the “maximum term of imprisonment” prescribed by law for an “offense” to have the same meaning in ACCA.
Our decision in United States v. R. L. C., 503 U. S. 291 (1992), is not to the contrary. The statutory provision there, 18 U. S. C. § 5037(c) (2000 ed.), set out the term of official detention for a juvenile found to be a delinquent. This provision was amended by the Sentencing Reform Act, see § 214, 98 Stat. 2013, and then amended again two years later, see §§21(a)(2)-(4), 100 Stat. 3596. As thus amended, the provision did not refer to the “maximum term of imprisonment” prescribed for an “offense.” Rather, the provision focused on the particular juvenile being sentenced. It provided that, “fin the case of a juvenile who is less than eighteen years old,’” official detention could not extend beyond the earlier of two dates: the juvenile’s 21st birthday or “‘the maximum term of imprisonment that would be authorized if the juvenile had been tried and convicted as an adult.’” United States v. R. L. C., supra, at 295-296, n. 1 (quoting 18 *393U. S. C. § 5037(c)). Because this provision clearly focuses on the circumstances of the particular juvenile and not on the offense, 503 U. S., at 299, it is not analogous to the ACCA provision that is before us in this case.
* * *
For these reasons, we hold that the “maximum term of imprisonment... prescribed by law” for the state drug convictions at issue in this case was the 10-year maximum set by the applicable recidivist provision. Accordingly, we reverse the judgment of the Court of Appeals and remand the case for further proceedings consistent with this opinion.
It is so ordered.
Justice Souter, with whom Justice Stevens and Justice Ginsburg join, dissenting.
The Court chooses one reading of the Armed Career Criminal Act (ACCA), 18 U. S. C. § 924(e) (2000 ed. and Supp. V), over another that would make at least as much sense of the statute’s ambiguous text and would follow the counsel of a tradition of lenity in construing perplexing criminal laws. The Court’s choice, moreover, promises hard times for the trial courts that will have to make the complex sentencing calculations this decision demands. I respectfully dissent.
I
The ACCA mandates a 15-year minimum sentence for anyone convicted of violating § 922(g) (2000 ed.) who “has three previous convictions [for] a serious drug offense” among his prior crimes. § 924(e)(1) (2000 ed., Supp. V). Section 924(e)(2)(A) (2000 ed.) defines “serious drug offense” as an offense under state or federal drug laws, “for which a maximum term of imprisonment of ten years or more is prescribed by law.” This limitation leaves open the question whether a given conviction qualifies as “serious” by refer*394ence to the penalty for the acts making up the basic offense, regardless of who commits it, or whether account must also be taken of further facts (such as an offender’s criminal record that qualified him for an enhanced penalty at the time of that earlier conviction). If the first alternative is the reading Congress intended, a sentencing judge needs to look only to the penalty specified for the basic offense committed by a first-time offender. But if the second is the intended one, a judge may have to consider sentencing variations (for using a gun, say, or for repeating the offense) set out in other provisions.
It all turns on the meaning of the word “offense,” to which the “maximum term” is tied. One can naturally read “an offense” at a general level as synonymous with “a crime,” which would tend to rule out reference to máximums adjusted for other facts; we do not usually speak of a crime of “burglary while having a criminal record and while out on bail.” Those details would come up only if we were speaking about a specific instance, described as a burglary “committed by someone with a record while out on bail,” in which case the other facts may “enhance” his sentence beyond what would have been the maximum term for burglary. The trouble is that “offense” could easily refer to a specific occurrence, too; looking at it that way would make it less jarring to suggest that the circumstances around an event that authorize higher penalty ranges (such as the use of a gun) or the defendant’s history (like a prior conviction) ought to count in identifying the maximum penalty for the offense committed on the given day, at the given place, by the particular offender, in a given way. Either reading seems to offer a plausible take on the “offense” for which the ACCA court will have to identify or calculate “maximum” penalties, under state law.
We get no help from imagining the circumstances in which a sentencing court would ask which reading to adopt. The choice of answer would be easy if the question arose in the *395mind of a lawyer whose client is thinking about a guilty plea and asks what maximum term he faces. See ante, at 383. His lawyer knows that he means the maximum term for him in his case. When a repeat offender wants to know, counsel understands that the penalty prescribed for the basic crime without the recidivist add-on is not the baseline for comparison that may make or break the potential plea agreement. And if the repeat offender faces a further statutory enhancement for carrying a gun during the offense, or for being out on bail, his lawyer would not tell him the maximum term for repeat offenders without guns or bail restrictions. By the same token, if the offender faced (as Rodriquez did) a lower sentence ceiling than what the statute says, by grace of mandatory sentencing guidelines, his lawyer would know enough to tell him that his maximum was capped in this way.
When the issue comes up not in a particular client’s questions about his own prospects, however, but in a trial judge’s mind wondering about the meaning of the general statute, context gives no ready answer. Nor does it break the tie to say, as the Court does, that taking “maximum” to refer to the basic offense would mean that a recidivist with add-ons could be sentenced above the ACCA “maximum,” see ante, at 383 (“[E]ven if respondent had been sentenced to, say, six years’ imprisonment, ‘the maximum term of imprisonment’ prescribed by law still would have been five years”). That description, after all, might be just a verbal quirk showing the statutory design in proper working order: if Congress meant an offense to be viewed generically and apart from offender characteristics, a gap between the maximum for ACCA purposes, and a heavier, actual sentence accounting for a defendant’s history is to be expected.1
*396The text does not point to any likelier interpretive choices, and as between these alternatives, it is simply ambiguous. 2 Because I do not believe its ambiguity is fairly resolved in the Government’s favor, I would affirm.
II
A
None of the Court’s three principal points or ripostes solves the puzzle. To begin with, there is something arbitrary about trying to resolve the ambiguity by rejecting the maximum-for-basic-offense option while declining to consider an entire class of offender-based sentencing adjustments. If offender characteristics are going to count in identifying the relevant maximum penalty, it would seem to follow that in jurisdictions with mandatory sentencing guidelines, the maximum “prescribed by law” would be what the guidelines de*397termine. The original Federal Guidelines, and the mandatory state guidelines I am aware of, were established under statutory authority that invests a guideline with the same legal status as a customary penalty provision. Cf. United States v. R. L. C., 503 U. S. 291, 297 (1992) (“The answer to any suggestion that the statutory character of a specific penalty provision gives it primacy over administrative sentencing guidelines is that the mandate to apply the Guidelines is itself statutory”).
The Court tries to deflect the implication of its position by denying that state sentencing guidelines really do set maximum penalties, since typically they allow a judge to depart from them, up or down, when specified conditions are met. See ante, at 390-391. But while this is true, the objection stands. However a particular mandatory guideline scheme works, it sets a maximum somewhere; if it includes conditions affecting what would otherwise be a guideline maximum, the top of the range as affected should be the relevant maximum on the Court’s reading of the statute. Indeed, the factual conditions involved are usually offender characteristics, and if the ACCA is going to count them under offense-defining statutes or freestanding recidivism laws, those same facts ought to count under a guideline rule (whether setting, or authorizing a departure from, a particular limit). There is no practical difference whether máximums are adjusted by a statute, a statutorily mandated guideline, or a guideline-specified departure; wherever a “prescription] by law” resides, it ought to be honored by the ACCA court.
If we were to follow the Court’s lights, then, I think we would have to accept the complication that guidelines schemes present, and face the difficulty of calculating enhanced máximums in guidelines jurisdictions.3 What we cannot do is resolve statutory ambiguity by looking to the *398sentencing range for an imaginary offender who meets statutory conditions for altering the basic sentence, but is artificially stripped of any characteristic that triggers a guideline rule also “prescribed by law.”
B
The more fundamental objection, though, goes to the Court’s basic conclusion that it makes the better sense to read the ACCA as resting the federal treatment of recidivists on the maximum sentence authorized by state recidivist schemes, in cases where state law must be considered. The Court says it would have been natural for Congress to think in terms of state judgments about repeat criminals when thinking about what to do at the national level, and the Court is quite possibly right about this; the fact that the federal penalty may turn on a state felony classification at all shows that Congress was thinking about state law. But the chances are at least equally good that the Court is wrong; it is odd to think that Congress would have piggybacked the federal system on state repeat-offender schemes, given the extraordinary and irreconcilable variations among state policies on the subject.
For one thing, the States’ recidivism schemes vary in their methods for augmenting sentences. Iowa’s law, for example, subjects repeat drug offenders to triple penalties, Iowa Code § 124.411(1) (2005); but in Wisconsin a repeat drug distributor will see his maximum term increased by a fixed number of years, whatever the starting point, see, e. g., Wis. Stat. § 961.48(l)(b) (2003-2004) (4-year increase for Class H felony such as selling one kilogram of marijuana).
More striking than differing structures, though, are the vast disparities in severity from State to State: under Massachusetts drug laws, a third conviction for selling a small amount of marijuana carries a maximum of 2.5 years. Mass. Gen. Laws, ch. 94C, §32C(b) (West 2006). In Delaware, a third conviction means a mandatory sentence of life in prison *399without parole. See Del. Code Aun., Tit. 11, § 4214(b) (2007) (third-felony penalty of life without parole for violations of nonnarcotic controlled substances law, Tit. 16, § 4752 (2003)). That Congress might have chosen to defer to state-law judgments about “seriousness” that vary so widely for the same conduct is at least open to doubt. And that doubt only gets worse when we notice that even where two States have similar maximum penalties for a base-level offense, their recidivist enhancements may lead the same conduct to trigger the ACCA sanction in one State but not the other: on the Court’s view, an offender’s second conviction for selling, say, just over two pounds of marijuana will qualify as an ACCA predicate crime if the conviction occurred in Arizona (maximum of 13 years), Iowa (15 years), Utah (15 years), and the District of Columbia (10 years), for example;4 but it will fall short of the mark in California (8 years), Michigan (8 years), and New York (8 years).5 Yet in each of these States, the base-level offense has a maximum term falling within a much narrower range (between 3.5 and 5.5 years).6 With this backdrop of state law, the Government can hardly be heard *400to say that there would be something “incongruous” about a federal law targeting offenses flagged by the penalties assigned only to bare conduct, without regard to recidivism or other offender facts. Brief for United States 17.
Nor does it show what the ACCA means by “maximum” or “offense” when the Court points to language from our prior cases saying that enhanced recidivist penalties are not to be viewed as retroactive punishment for past crimes, for purposes of double-jeopardy and right-to-counsel enquiries. See ante, at 385-386 (citing Nichols v. United States, 511 U. S. 738, 747 (1994), and Gryger v. Burke, 334 U. S. 728, 732 (1948)). The quotations show that a separate offense is identified by an enhanced penalty, the Court says, because from them we can draw the conclusion that “[w]hen a defendant is given a higher sentence under a recidivism statute,” nonetheless “100% of the punishment is for the offense of conviction,” leaving nothing to be attributed to “prior convictions or the defendant’s ‘status as a recidivist,’ ” ante, at 386.
Still, the fact is that state-law máximums for repeat offenders sometimes bear hardly any relation to the gravity of the triggering offense, as “three-strikes” laws (not to mention the Delaware example, above) often show. See, e. g., Ill. Comp. Stat., ch. 720, §5/33B-l (2004) (mandatory life sentence for third “Class X” felony, such as dealing heroin, without regard to the specific penalty gradation for the latest Class X felony or to any similarity with prior offenses); W. Va. Code Ann. § 61-ll-18(c) (2005) (if offender was “twice before convicted in the United States of a crime punishable by confinement in a penitentiary,” third such conviction incurs a mandatory life sentence). Cf. Ewing v. California, 538 U. S. 11,30, n. 2 (2003) (plurality opinion) (the “California Legislature therefore made a deliberate policy decision . . . that the gravity of the new felony should not be a determinative factor in triggering the application of the Three Strikes Law” (internal quotation marks omitted)). And there is no *401denying that the fact of prior convictions (or a defendant’s recidivist status) is necessary for the “ ‘stiffened penalty’ ” to be imposed for “ ‘the latest crime,’ ” ante, at 386, the necessary fact being specific to the offender, and falling outside the definition of the offense. This is, after all, what it means to apply an “enhancement.”
The upshot is that it may have been natural for Congress to think of state recidivism schemes, but it may well not have been. If there is anything strange about ignoring enhanced penalties, there is something at least as strange about a federal recidivist statute that piles enhancement on enhancement, magnifying the severity of state laws severe to begin with.
C
Whatever may be the plausibility of the offender-based reading of the statute as the Court describes it, the Court’s description avoids a source of serious doubt by glossing over the practical problems its take on the statute portends. The Court is unmoved by the argument that Congress probably did not expect federal courts applying the ACCA to master the countless complications of state sentencing schemes; because all jurisdictions provide for enhanced sentencing some way or another, the Court thinks there is nothing threatening in the subject, which it tries to simplify by offering a few practical pointers. It notes that there will be cases with a qualifying enhancement “evident from the length of the sentence imposed” by the state court; sometimes, it says, a court’s “judgment of conviction will . . . list the maximum possible sentence”; or the state prosecutor will have “submit[ted] a formal charging document in order to obtain a recidivist enhancement.” Ante, at 389. And in cases involving pleas, the Court notes, “the plea colloquy will very often include a statement by the trial judge regarding the maximum penalty.” Ibid. Even when there are no pointers to help, says the Court, and “the records that may properly be *402consulted” yield no clear answer, the worst that can happen will be the Government’s inability to show that a prior conviction qualifies. Ante, at 389.
But it is not that easy, and the Court’s pointers are not much comfort. To start with, even where a “maximum” sentence is mentioned in state records, how will the ACCA court be supposed to know that the “maximum” written down there is what the Court today holds that “maximum” means? A State’s number below 10 years may refer to the base-level offense, or it may be the reduced maximum required by mandatory guidelines; and a number over 10 years may be the product of other enhancements (as for weapons use or being out on bail at the time of commission). Having to enquire into just what imposed sentences or what trial documents really mean would seem to leave plenty of sorting out for the federal courts to do (or at least, for federal prosecutors, if they end up with the job).
Another example: state laws are not written to coordinate with the ACCA, and if a State’s specific repeat drug-offender provisions, say, are supposed to be read together with its general habitual-offender statutes, the resulting “maximum” may not be the Court’s “maximum.” Indeed, a federal court may have to figure out just how those state statutes may be read together to avoid conflict between them, when the way to avoid conflict is not clear cut even for the state courts, see, e. g., Goldberg v. State, 282 Ga. 542, 651 S. E. 2d 667 (2007) (general recidivist statute trumps more specific one; overruling same court’s decision in Mikell v. State, 270 Ga. 467,510 S. E. 2d 523 (1999)); State v. Keith, 102 N. M. 462,697 P. 2d 145 (App. 1985) (specific trumps general). Cf. Clines v. State, 912 So. 2d 550 (Fla. 2005) (relying on rule of lenity to resolve whether multiple recidivist categories in same habitual-offender law could apply to a single sentence).
And there is more: as Rodriquez reminds us, just deciding what counts as a “prior” offense under state law is not always an easy thing. See People v. Wiley, 9 Cal. 4th 580, *403583, 889 P. 2d 541, 542 (1995). (noting difficulty of applying requirement that “prior” charges have been “brought and tried separately,” where defendant had been convicted in trials occurring one day apart and sentenced at the same court session; in the end, drawing the needed inference from docket numbers revealed on documents requested from the municipal trial court); id., at 595,889 P. 2d, at 550 (Werdegar, J., dissenting) (protesting the court’s solicitation and use of extrarecord documents). Nor would that sort of enquiry get any easier, or be more likely to benefit from well-settled state law, when a given State’s law takes account of prior offenses in other States, see Timothy v. State, 90 P. 3d 177 (Alaska App. 2004) (holding Oklahoma burglary not to be analogous to one in Alaska, for purposes of Alaska’s recidivism enhancements, thus overruling its own 2-year-old decision, Butts v. State, 53 P. 3d 609 (2002)); or, to take a specific example, when what qualifies a prior offense under one State’s recidivism scheme is the length of the sentence authorized by another State’s law (raising the question whether that first State would see recidivist enhancements the same way the Court does today). See, e.g., N. J. Stat. Ann. § 2C:44-4(c) (West 2005) (“A conviction in another jurisdiction shall constitute a prior conviction of a crime if a sentence of imprisonment in excess of 6 months was authorized under the law of the other jurisdiction”); N. M. Stat. Ann. §31-18-17(D)(2)(b) (2007 Supp.) (defining “prior felony conviction” as, inter alia, a felony “punishable [by] a maximum term of imprisonment of more than one year”).
A still thornier problem is how federal courts are supposed to treat a State’s procedural safeguards for using prior convictions at sentencing. Saying that congressional deference to the States’ judgments about the severity of crimes also extends to their judgments about recidivism raises, but does not answer, the question whether such deference goes only as far as the state courts themselves could go in raising penalties. (The Court’s disregard of mandatory sentencing *404guidelines would seem to suggest that the answer is no.) In those States that require notice before the prosecutor can seek a recidivism enhancement, for example, how will a federal court decide whether the ACCA counts a prior conviction that would have qualified for recidivism enhancement if the state prosecutor had not failed to give timely notice? See, e. g., Commonwealth v. Fernandes, 430 Mass. 517, 522, 722 N. E. 2d 406, 409 (1999) (noting longstanding rule that the indictment must give notice of prior convictions “that may subject the defendant to enhanced punishment”).
I could go on, but this is enough to show that the Court’s interpretation promises that ACCA courts will face highly complicated enquiries into every State’s or Territory’s collection of ancillary sentencing laws. That is an unconvincing answer to the ambiguity.
Ill
At the end of the day, a plainly superior reading may well be elusive; one favoring the Government certainly is. It does not defy common English or common sense, after all, to look at a statute with one penalty range for the basic crime and a higher one for a repeat offender and say that the former sets the maximum penalty for the “offense”; but neither is it foolish to see the “offense” as defined by its penalty, however that is computed. What I have said so far suggests that I think the basic-crime view of “offense” is the better one, but I will concede that the competing positions are pretty close to evenly matched. And on that assumption, there is a ready tiebreaker.
The interpretation adopted by both the District Court and the Court of Appeals is the one counseled by the rule of lenity, which applies where (as here) we have “ ‘seiz[ed] every thing from which aid can be derived,’” but are “left with an ambiguous statute,” United States v. Bass, 404 U. S. 336, 347 (1971) (quoting United States v. Fisher, 2 Cranch 358, 386 (1805) (opinion for the Court by Marshall, C. J.)). The rule is grounded in “ ‘the instinctive distaste against men languish*405ing in prison unless the lawmaker has clearly said they should/” Bass, supra, at 348 (quoting H. Friendly, Benchmarks 209 (1967)), and we have used it to resolve questions both about metes and bounds of criminal conduct and about the severity of sentencing, see Bifulco v. United States, 447 U. S. 381, 387 (1980) (collecting cases). “This policy of lenity means that the Court will not interpret a federal criminal statute so as to increase the penalty that it places on an individual when such an interpretation can be based on no more than a guess as to what Congress intended.” Ladner v. United States, 358 U. S. 169, 178 (1958).
This is why lenity should control here. Even recognizing the best that can be said for the Government’s side, its position rests on debatable guesswork to send a man to prison for 180 months, as against 92 months on the basic-crime view. And the District Courts will be imposing higher sentences more than doubling the length of the alternative in a good many other cases, as well.
The “fair warning” that motivates the lenity rule, McBoyle v. United States, 283 U. S. 25,27 (1931) (opinion for the Court by Holmes, J.), may sometimes be a benign fiction, see R. L. C., 503 U. S., at 309 (SCALIA, J., concurring in part and concurring in judgment), but there is only one reading of this statute with any realistic chance of giving fair notice of how the ACCA will apply, and that is the reading the District Court and the Court of Appeals each chose. Their choice should be ours, too.