James v. United States 550 U.S. 192, 167 L. Ed. 2d 532, 127 S. Ct. 1586, 2007 U.S. LEXIS 4337, SCDB 2006-032

Justice Alito

delivered the opinion of the Court.

The Armed Career Criminal Act (ACCA), 18 U. S. C. § 924(e)(1) (2000 ed., Supp. IV), provides that a defendant convicted of possession of a firearm by a convicted felon, in violation of § 922(g), is subject to a mandatory sentence of 15 years of imprisonment if the defendant has three prior convictions “for a violent felony or a serious drug offense.” The question before us is whether attempted burglary, as defined by Florida law, is a “violent felony” under ACCA. We hold that it is, and we therefore affirm the judgment of the Court of Appeals.

I

Petitioner Alphonso James pleaded guilty in federal court to one count of possessing a firearm after being convicted of a felony, in violation of § 922(g)(1). In his guilty plea, James *196admitted to the three prior felony convictions listed in his federal indictment. These included a conviction in Florida state court for attempted burglary of a dwelling, in violation of Fla. Stat. §§810.02 and 777.04 (1993).1

At sentencing, the Government argued that James was subject to ACCA’s 15-year mandatory minimum term because of his three prior convictions. James objected, arguing that his attempted burglary conviction did not qualify as a “violent felony” under 18 U. S. C. § 924(e). The District Court held that attempted burglary is a violent felony, and the Court of Appeals for the Eleventh Circuit affirmed that holding, 430 F. 3d 1150, 1157 (2005). We granted certiorari, 547 U. S. 1191 (2006).

II

A

ACCA’s 15-year mandatory minimum applies “[i]n the case of a person who violates section 922(g) of this title [the felon in possession of a firearm provision] and has three previous convictions ... for a violent felony or a serious drug offense, or both, committed on occasions different from one another.” § 924(e)(1) (2000 ed., Supp. IV). ACCA defines a “violent felony” as

“any crime punishable by imprisonment for a term exceeding one year ... that—

“(i) has as an element the use, attempted use, or threatened use of physical force against the person of another; or

“(ii) is burglary, arson, or extortion, involves use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.” § 924(e)(2)(B).

*197Florida defined the crime of burglary at the time of James’ conviction as follows: “ ‘Burglary’ means entering or remaining in a structure or a conveyance with the intent to commit an offense therein, unless the premises are at the time open to the public or the defendant is licensed or invited to enter or remain.” Fla. Stat. §810.02(1). Florida’s criminal attempt statute provided: “A person who attempts to commit an offense prohibited by law and in such attempt does any act toward the commission of such offense, but fails in the perpetration or is intercepted or prevented in the execution thereof, commits the offense of criminal attempt.” § 777.04(1). The attempted burglary conviction at issue here was punishable by imprisonment for a term exceeding one year.

The parties agree that attempted burglary does not qualify as a “violent felony” under clause (i) of ACCA’s definition because it does not have “as an element the use, attempted use, or threatened use of physical force against the person of another.” 18 U. S. C. § 924(e)(2)(B)(i). Nor does it qualify as one of the specific crimes enumerated in clause (ii). Attempted burglary is not arson or extortion. It does not involve the use of explosives. And it is not “burglary” because it does not meet the definition of burglary under ACCA that this Court set forth in Taylor v. United States, 495 U. S. 575, 598 (1990): “an unlawful or unprivileged entry into, or remaining in, a building or other structure, with intent to commit a crime.” See Fla. Stat. § 777.04(1) (crime of attempt under Florida law requires as an element that the defendant “fai[l] in the perpetration or [be] intercepted or prevented in the execution” of the underlying offense).

The question before the Court, then, is whether attempted burglary, as defined by Florida law, falls within ACCA’s residual provision for crimes that “otherwise involv[e] conduct that presents a serious potential risk of physical injury to another.” 18 U. S. C. § 924(e)(2)(B)(ii).

*198B

Before determining whether the elements of attempted burglary under Florida law qualify under ACCA’s residual provision, we first consider James’ argument that the statute’s text and structure categorically exclude attempt offenses from the scope of the residual provision. We conclude that nothing in the plain language of clause (ii), when read together with the rest of the statute, prohibits attempt offenses from qualifying as ACCA predicates when they involve conduct that presents a serious potential risk of physical injury to another.

James first argues that the residual provision of clause (ii) must be read in conjunction with clause (i), which expressly includes in its definition of “violent felony” offenses that have “as an element the... attempted use... of physical force against the person of another.” § 924(e)(2)(B)(i) (emphasis added). James thus concludes that Congress’ express inclusion of attempt offenses in clause (i), combined with its failure to mention attempts in clause (ii), demonstrates an intent to categorically exclude attempt offenses from the latter provision.

We are not persuaded. James’ reading would unduly narrow clause (ii)’s residual provision, the language of which does not suggest any intent to exclude attempt offenses that otherwise meet the statutory criteria. Clause (i), in contrast, lacks a broad residual provision, thus making it necessary to specify exactly what types of offenses — including attempt offenses — are covered by its language. In short, “the expansive phrasing of” clause (ii) “points directly away from the sort of exclusive specification” that James would read into it. Chevron U S. A. Inc. v. Echazabal, 536 U. S. 73, 80 (2002); see also United States v. Davis, 16 F. 3d 212, 217 (CA7) (rejecting argument that “had Congress wished to include attempted burglary as a § 924(e) predicate offense, it would have done so expressly” as “untenable in light of the very existence of the ‘otherwise’ clause, which Congress *199plainly included to serve as a catch-all provision”), cert, denied, 513 U. S. 945 (1994).

James next invokes the canon of ejusdem generis — that when a general phrase follows a list of specifies, it should be read to include only things of the same type as those specifically enumerated. He argues that the “common attribute” of the offenses specifically enumerated in clause (ii) — burglary, arson, extortion, and crimes involving the use of explosives — is that they are all completed offenses. The residual provision, he contends, should similarly be read to extend only to completed offenses.

This argument is unavailing. As an initial matter, the premise on which it depends — that clause (ii)’s specifically enumerated crimes are limited to completed offenses — is false. An unsuccessful attempt to blow up a government building, for example, would qualify as a specifically enumerated predicate offense because it would “involv[e] [the] use of explosives.” See, e.g., § 844(f)(1) (2000 ed., Supp. IV) (making it a crime to “maliciously damag[e] or destro[y], or attempft] to damage or destroy, by means of fire or an explosive,” certain property used in or affecting interstate commerce (emphasis added)).

In any event, the most relevant common attribute of the enumerated offenses of burglary, arson, extortion, and explosives use is not “completion.” Rather, it is that all of these offenses, while not technically crimes against the person, nevertheless create significant risks of bodily injury or confrontation that might result in bodily injury. As we noted in Taylor:

“Congress thought that certain general categories of property crimes — namely burglary, arson, extortion, and the use of explosives — so often presented a risk of injury to persons, or were so often committed by career criminals, that they should be included in the enhancement statute even though, considered solely in terms of their statutory elements, they do not necessarily involve the *200use or threat of force against a person.” 495 U. S., at 597.

See also id., at 588 (noting that Congress singled out burglary because it “often creates the possibility of a violent confrontation”); United States v. Adams, 51 Fed. Appx. 507, 508 (CA6 2002) (arson presents “a serious potential risk of physical injury to another” because “[n]ot only might the targeted building be occupied,” but also “the fire could harm firefighters and onlookers and could spread to occupied structures”); H. R. Rep. No. 99-849, p. 3 (1986) (purpose of clause (ii) was to “add State and Federal crimes against property such as burglary, arson, extortion, use of explosives and similar crimes as predicate offenses where the conduct involved presents a serious risk of injury to a person”).

Congress’ inclusion of a broad residual provision in clause (ii) indicates that it did not intend the preceding enumerated offenses to be an exhaustive list of the types of crimes that might present a serious risk of injury to others and therefore merit status as a § 924(e) predicate offense. Nothing in the statutory language supports the view that Congress intended to limit this category solely to completed offenses.

C

James also relies on ACCA’s legislative history to buttress his argument that clause (ii) categorically excludes attempt offenses. In the deliberations leading up to ACCA’s adoption in 1984, the House rejected a version of the statute that would have provided enhanced penalties for use of a firearm by persons with two prior convictions for “any robbery or burglary offense, or a conspiracy or attempt to commit such an offense.” S. 52, 98th Cong., 2d Sess., § 2 (1984) (emphasis added). The bill that ultimately became law omitted any reference to attempts, and simply defined “violent felony” to include “robbery or burglary, or both.” Armed Career Criminal Act of 1984, § 1802, 98 Stat. 2185, repealed in 1986 by Pub. L. 99-308, § 104(b), 100 Stat. 459. James argues *201that Congress’ rejection of this explicit “attempt” language in 1984 evidenced an intent to exclude attempted burglary as a predicate offense.

Whatever weight this legislative history might ordinarily have, we do not find it probative here, because the 1984 enactment on which James relies was not Congress’ last word on the subject. In 1986, Congress amended ACCA for the purpose of “‘expanding’ the range of predicate offenses.” Taylor, supra, at 584. The 1986 amendments added the more expansive language that is at issue in this case — including clause (ii)’s language defining as violent felonies offenses that are “burglary, arson, or extortion, involv[e] use of explosives, or otherwise involv[e] conduct that presents a serious potential risk of physical injury to another.” Career Criminals Amendment Act of 1986, § 1402(b), 100 Stat. 3207-40, codified at 18 U.S.C. § 924(e)(2)(B)(ii). This language is substantially broader than the 1984 provision that it amended. Because both the Government and the Court of Appeals relied on the broader language of the 1986 amendments — specifically, the residual provision — as the textual basis for including attempted burglary within the law’s scope, Congress’ rejection of express language including attempt offenses in the 1984 provision is not dispositive. Congress did not consider, much less reject, any such language when it enacted the 1986 amendments. What it did consider, and ultimately adopted, was a broadly worded residual clause that does not by its terms exclude attempt offenses, and whose reach is broad enough to encompass at least some such offenses.

Ill

Having concluded that neither the statutory text nor the legislative history discloses any congressional intent to categorically exclude attempt offenses from the scope of § 924(e)(2)(B)(ii)’s residual provision, we next ask whether attempted burglary, as defined by Florida law, is an offense that “involves conduct that presents a serious potential risk *202of physical injury to another.” In answering this question, we employ the “ ‘categorical approach’ ” that this Court has taken with respect to other offenses under ACCA. Under this approach, we “Took only to the fact of conviction and the statutory definition of the prior offense,’” and do not generally consider the “particular facts disclosed by the record of conviction.” Shepard v. United States, 544 U. S. 13, 17 (2005) (quoting Taylor, 495 U. S., at 602). That is, we consider whether the elements of the offense are of the type that would justify its inclusion within the residual provision, without inquiring into the specific conduct of this particular offender.

A

We begin by examining what constitutes attempted burglary under Florida law. On its face, Florida’s attempt statute requires only that a defendant take “any act toward the commission” of burglary. Fla. Stat. § 777.04(1). James contends that this broad statutory language sweeps in merely preparatory activity that poses no real danger of harm to others — for example, acquiring burglars’ tools or casing a structure while planning a burglary.

But while the statutory language is broad, the Florida Supreme Court has considerably narrowed its application in the context of attempted burglary, requiring an “overt act directed toward entering or remaining in a structure or conveyance.” Jones v. State, 608 So. 2d 797, 799 (1992). Mere preparation is not enough. See ibid.2 Florida’s lower *203courts appear to have consistently applied this heightened standard. See, e. g., Richardson v. State, 922 So. 2d 331, 334 (App. 2006); Davis v. State, 741 So. 2d 1213, 1214 (App. 1999).

The pivotal question, then, is whether overt conduct directed toward unlawfully entering or remaining in a dwelling, with the intent to commit a felony therein, is “conduct that presents a serious potential risk of physical injury to another.” 18 U. S. C. § 924(e)(2)(B)(ii).

B

In answering this question, we look to the statutory language for guidance. The specific offenses enumerated in clause (ii) provide one baseline from which to measure whether other similar conduct “otherwise ... presents a serious potential risk of physical injury.” In this case, we can ask whether the risk posed by attempted burglary is comparable to that posed by its closest analog among the enumerated offenses — here, completed burglary. See Taylor, supra, at 600, n. 9 (“The Government remains free to argue that any offense — including offenses similar to generic burglary — should count towards enhancement as one that ‘otherwise involves conduct that presents a serious potential risk of physical injury to another’ under § 924(e)(2)(B)(ii)”).

The main risk of burglary arises not from the simple physical act of wrongfully entering onto another’s property, but rather from the possibility of a face-to-face confrontation between the burglar and a third party — whether an occupant, a police officer, or a bystander — who comes to investigate. That is, the risk arises not from the completion of the burglary, but from the possibility that an innocent person might appear while the crime is in progress.

Attempted burglary poses the same kind of risk. Interrupting an intruder at the doorstep while the would-be burglar is attempting a break-in creates a risk of violent *204confrontation comparable to that posed by finding him inside the structure itself. As one court has explained:

“In all of these cases the risk of injury arises, not from the completion of the break-in, but rather from the possibility that some innocent party may appear on the scene while the break-in is occurring. This is just as likely to happen before the defendant succeeds in breaking in as after. Indeed, the possibility may be at its peak while the defendant is still outside trying to break in, as that is when he is likely to be making noise and exposed to the public view.. .. [T]here is a serious risk of confrontation while a perpetrator is attempting to enter the building.” United States v. Payne, 966 F. 2d 4, 8 (CA1 1992).

Indeed, the risk posed by an attempted burglary that can serve as the basis for an ACCA enhancement may be even greater than that posed by a typical completed burglary. All burglaries begin as attempted burglaries. But ACCA only concerns that subset of attempted burglaries where the offender has been apprehended, prosecuted, and convicted. This will typically occur when the attempt is thwarted by some outside intervenor — be it a property owner or law enforcement officer. Many completed burglaries do not involve such confrontations. But attempted burglaries often do; indeed, it is often just such outside intervention that prevents the attempt from ripening into completion.

Concluding that attempted burglary presents a risk that is comparable to the risk posed by the completed offense, every Court of Appeals that has construed an attempted burglary law similar in scope to Florida’s has held that the offense qualifies as a “violent felony” under clause (ii)’s residual provision.3 The only cases holding to the contrary *205involved attempt laws that could be satisfied by preparatory conduct that does not pose the same risk of violent confrontation and physical harm posed by an attempt to enter a structure illegally.4 Given that Florida law, as interpreted *206by that State’s highest court, requires an overt act directed toward the entry of a structure, we need not consider whether the more attenuated conduct encompassed by such laws presents a potential risk of serious injury under ACCA.

The United States Sentencing Commission has come to a similar conclusion with regard to the Sentencing Guidelines’ career offender enhancement, whose definition of a predicate “crime of violence” closely tracks ACCA’s definition of “violent felony.” See United States Sentencing Commission, Guidelines Manual §4B1.2(a)(2) (Nov. 2006) (USSG). The Commission has determined that “crime[s] of violence” for the purpose of the Guidelines enhancement “include the offenses of aiding and abetting, conspiring, and attempting to commit such offenses.” §4B1.2, comment., n. 1. This judgment was based on the Commission’s review of empirical sentencing data and presumably reflects an assessment that attempt crimes often pose a similar risk of injury as completed offenses. As then-Chief Judge Breyer explained, “[t]he Commission, which collects detailed sentencing data on virtually every federal criminal case, is better able than any individual court to make an informed judgment about the relation between” a particular offense and “the likelihood of accompanying violence.” United States v. Doe, 960 F. 2d 221, 225 (CA1 1992); see also USSG §1A3 (Nov. 1987), re*207printed in § 1 A1.1 comment. (Nov. 2006) (describing empirical basis of Commission’s formulation of Guidelines); United States v. Chambers, 478 F. 3d 724 (CA7 2007) (noting the usefulness of empirical analysis from the Commission in determining whether an unenumerated crime poses a risk of violence). While we are not bound by the Sentencing Commission’s conclusion, we view it as further evidence that a crime like attempted burglary poses a risk of violence similar to that presented by the completed offense.

C

James responds that it is not enough that attempted burglary “ ‘generally’ ” or in “ ‘most cases’ ” will create a risk of physical injury to others. Brief for Petitioner 32. Citing the categorical approach we employed in Taylor, he argues that we cannot treat attempted burglary as an ACCA predicate offense unless all cases present such a risk. James’ approach is supported by neither the statute’s text nor this Court’s holding in Taylor.

One could, of course, imagine a situation in which attempted burglary might not pose a realistic risk of confrontation or injury to anyone — for example, a break-in of an unoccupied structure located far off the beaten path and away from any potential intervenors. But ACCA does not require metaphysical certainty. Rather, § 924(e)(2)(B)(ii)’s residual provision speaks in terms of a “potential risk.” These are inherently probabilistic concepts.5 Indeed, the combination of the two terms suggests that Congress intended to encompass possibilities even more contingent or remote than a sim*208pie “risk,” much less a certainty. While there may be some attempted burglaries that do not present a serious potential risk of physical injury to another, the same is true of completed burglaries — which are explicitly covered by the statutory language and provide a baseline against which to measure the degree of risk that a nonenumerated offense must “otherwise” present in order to qualify.

James’ argument also misapprehends Taylor's categorical approach. We do not view that approach as requiring that every conceivable factual offense covered by a statute must necessarily present a serious potential risk of injury before the offense can be deemed a violent felony. Cf. Gonzales v. Duenas-Alvarez, 549 U. S. 183, 193 (2007) (“[T]o find that a state statute creates a crime outside the generic definition of a listed crime in a federal statute requires more than the application of legal imagination to a state statute’s language. It requires a realistic probability, not a theoretical possibility, that the State would apply its statute to conduct that falls outside the generic definition of a crime”).

Rather, the proper inquiry is whether the conduct encompassed by the elements of the offense, in the ordinary case, presents a serious potential risk of injury to another. One can always hypothesize unusual cases in which even a prototypically violent crime might not present a genuine risk of injury — for example, an attempted murder where the gun, unbeknownst to the shooter, had no bullets, see United States v. Thomas, 361 F. 3d 653, 659 (CADC 2004). Or, to take an example from the offenses specifically enumerated in § 924(e)(2)(B)(ii), one could imagine an extortion scheme where an anonymous blackmailer threatens to release embarrassing personal information about the victim unless he is mailed regular payments. In both cases, the risk of physical injury to another approaches zero. But that does not mean that the offenses of attempted murder or extortion are categorically nonviolent.

*209As long as an offense is of a type that, by its nature, presents a serious potential risk of injury to another, it satisfies the requirements of § 924(e)(2)(B)(ii)’s residual provision. Attempted burglary under Florida law — as construed in Jones to require an overt act directed toward entry of a structure — satisfies this test.

D

Justice Scalia’s dissent criticizes our approach on the ground that it does not provide sufficient guidance for lower courts required to decide whether unenumerated offenses other than attempted burglary qualify as violent felonies under ACCA. But the dissent’s alternative approach has more serious disadvantages. Among other things, that approach unnecessarily decides an important question that the parties have not briefed (the meaning of the term “extortion” in § 924(e)(2)(B)(ii)), decides that question in a way that is hardly free from doubt, and fails to provide an interpretation of the residual provision that furnishes clear guidance for future cases.

The dissent interprets the residual provision to require at least as much risk as the least dangerous enumerated offense. But the ordinary meaning of the language of the residual clause does not impose such a requirement. What the clause demands is “a serious potential risk of physical injury to another.” While it may be reasonable to infer that the risks presented by the enumerated offenses involve a risk of this magnitude, it does not follow that an offense that presents a lesser risk necessarily fails to qualify. . Nothing in the language of § 924(e)(2)(B)(ii) rules out the possibility that an offense may present “a serious risk of physical injury to another” without presenting as great a risk as any of the enumerated offenses.

Moreover, even if an unenumerated offense could not qualify without presenting at least as much risk as the least risky *210of the enumerated offenses, it would not be necessary to identify the least risky of those offenses in order to decide this case. Rather, it would be sufficient to establish simply that the unenumerated offense presented at least as much risk as one of the enumerated offenses. Thus, Justice Scalia’s interpretation of the meaning of the term “extortion” is unnecessary — and inadvisable. The parties have not briefed this issue, and the proposed interpretation is hardly beyond question. Instead of interpreting the meaning of the term “extortion” in accordance with its meaning at common law or in modern federal and state statutes, see Taylor, 495 U. S., at 598, it is suggested that we adopt an interpretation that seems to be entirely novel and that greatly reduces the reach of ACCA.

The stated reason for tackling this question is to provide guidance for the lower courts in future cases — surely a worthy objective. But in practical terms, the proposed interpretation of the residual clause would not make it much easier for the lower courts to decide whether other unenumerated offenses qualify. Without hard statistics— and no such statistics have been called to our attention— how is a lower court to determine whether the risk posed by generic burglary is greater or less than the risk posed by an entirely unrelated unenumerated offense — say, escape from prison?6

*211In the end, Justice Scalia’s analysis of this case turns on the same question as ours — i. e., the comparative risks presented by burglary and attempted burglary. The risk of physical injury in both cases occurs when there is a confrontation between the criminal and another person, whether an occupant of the structure, a law enforcement officer or security guard, or someone else. It is argued that when such an encounter occurs during a consummated burglary (i. e., after entry), the risk is greater than it is when the encounter occurs during an attempted burglary (i. e., before entry is effected), and that may be true. But this argument fails to come to grips with the fact that such encounters may occur much more frequently during attempted burglaries because it is precisely due to such encounters that many planned burglaries do not progress beyond the attempt stage. Justice Scalia dismisses the danger involved when encounters occur during attempted burglaries, stating that such encounters are “likely to consist of nothing more than the occupant’s yelling Who’s there?’ from his window, and the burglar’s running away.” Post, at 226. But there are many other possible scenarios. An armed would-be burglar may be spotted by a police officer, a private security guard, or a participant in a neighborhood watch program. Or a homeowner angered by the sort of conduct recited in James’ presentence report — throwing a hammer through a window — may give chase, and a violent encounter may ensue. For these rea*212sons and the reasons discussed above, we are convinced that the offense of attempted burglary, as defined by Florida law, qualifies under ACCA’s residual clause.

IV

Although the question on which this Court granted certiorari focused on the attempt prong of Florida’s attempted burglary law, James also argues that the scope of the State’s underlying burglary statute itself precludes treating attempted burglary as a violent felony for ACCA purposes. Specifically, he argues that Florida’s burglary statute differs from “generic” burglary as defined in Taylor, supra, at 598, because it defines a “‘[djwelling’” to include not only the structure itself, but also the “curtilage thereof,”7 Fla. Stat. §810.011(2).

We agree that the inclusion of curtilage takes Florida’s underlying offense of burglary outside the definition of “generic burglary” set forth in Taylor, which requires an unlawful entry into, or remaining in, “a building or other structure” 495 U. S., at 598 (emphasis added). But that conclusion is not dispositive, because the Government does not argue that James’ conviction for attempted burglary constitutes “burglary” under § 924(e)(2)(B)(ii). Rather, it relies on the residual provision of that clause, which — as the Court has recognized — can cover conduct that is outside the strict definition of, but nevertheless similar to, generic burglary. Id., at 600, n. 9.

Is the risk posed by an attempted entry of the curtilage comparable to that posed by the attempted entry of a structure (which, as we concluded above, is sufficient to qualify *213under the residual provision)? We must again turn to state law in order to answer this question.

The Florida Supreme Court has construed curtilage narrowly, requiring “some form of an enclosure in order for the area surrounding a residence to be considered part of the ‘curtilage’ as referred to in the burglary statute.” State v. Hamilton, 660 So. 2d 1038, 1044 (1995) (holding that a yard surrounded by trees was not “curtilage”); see also United States v. Matthews, 466 F. 3d 1271, 1274 (CA11 2006) (“Florida case law construes curtilage narrowly, to include only an enclosed area surrounding a structure”). Given this narrow definition, we do not believe that the inclusion of curtilage so mitigates the risk presented by attempted burglary as to take the offense outside the scope of clause (ii)’s residual provision.

A typical reason for enclosing the curtilage adjacent to a structure is to keep out unwanted visitors — especially those with criminal motives. And a burglar who illegally attempts to enter the enclosed area surrounding a dwelling creates much the same risk of physical confrontation with a property owner, law enforcement official, or other third party as does one who attempts to enter the structure itself. In light of Florida’s narrow definition of curtilage, attempted burglary of the curtilage requires both physical proximity to the structure and an overt act directed toward breaching the enclosure. Such an attempt “presents a serious potential risk that violence will ensue and someone will be injured.” Id., at 1275 (holding that burglary of the curtilage is a violent felony under ACCA’s residual provision).

V

Finally, James argues that construing attempted burglary as a violent felony raises Sixth Amendment issues under Apprendi v. New Jersey, 530 U. S. 466 (2000), and its progeny because it is based on “judicial fact finding” about the risk presented by “the acts that underlie ‘most’ convictions for *214attempted burglary.” Brief for Petitioner 34, 35. This argument is without merit.

In determining whether attempted burglary under Florida law qualifies as a violent felony under § 924(e)(2)(B)(ii), the Court is engaging in statutory interpretation, not judicial factfinding. Indeed, by applying Taylor’s categorical approach, we have avoided any inquiry into the underlying facts of James’ particular offense, and have looked solely to the elements of attempted burglary as defined by Florida law. Such analysis raises no Sixth Amendment issue.8

* * *

For these reasons, the judgment of the Court of Appeals for the Eleventh Circuit is affirmed.

It is so ordered.

Justice Scalia,

with whom

Justice Stevens and Justice Ginsburg join, dissenting.

I disagree with the Court’s basic approach in this case, and must therefore lay out my own.

I

As the Court acknowledges, ante, at 197, the only way attempted burglary can qualify as a violent felony under the Armed Career Criminal Act (ACCA) is by falling within the “residual provision” of clause (ii) — that is, if it is a crime that “otherwise involves conduct that presents a serious potential risk of physical injury to another.” 18 U. S. C. § 924(e)(2)(B)(ii). This was the basis for the Eleventh Circuit’s decision (“We ... hold that an attempt to commit bur*215glary . . . presents the potential risk of physical injury to another sufficient to satisfy the ACCA’s definition of a ‘violent felony,’ ” 430 F. 3d 1150, 1157 (2005)), and it is the center of the parties’ dispute before this Court.

The problem with the Court’s approach to determining which crimes fit within the residual provision is that it is almost entirely ad hoc. This crime, the Court says, does “involv[e] conduct that presents a serious potential risk of physical injury to another.” That gets this case off our docket, sure enough. But it utterly fails to do what this Court is supposed to do: provide guidance concrete enough to ensure that the ACCA residual provision will be applied with an acceptable degree of consistency by the hundreds of district judges who impose sentences every day. The one guideline the Court does suggest is that the sentencer should compare the unenumerated offense at issue with the “closest analog” among the four offenses that are set forth (burglary, arson, extortion, and crimes involving the use of explosives), and should include the unenumerated offense within ACCA if the risk it poses is “comparable.” Ante, at 203. The principal attraction of this test, I suspect, is that it makes it relatively easy to decide the present case (though, as I shall subsequently discuss, I think the Court reaches the wrong conclusion as to whether attempted burglary poses a comparable risk). Assuming that “comparable” means “about the same,” the Court’s test does provide some guidance where the most closely analogous offense is clear (as here) and the risk is comparable. But what if, as will very often be the case, it is not at all obvious which of the four enumerated offenses is the closest analog — or if (to tell the truth) none of them is analogous at all? Is, for example, driving under the influence of alcohol more analogous to burglary, arson, extortion, or a crime involving use of explosives? And if an analog is identified, what is to be done if the offense at issue does not present a comparable risk? The Court declines to say, but it seems inconceivable that it means the offense to *216be excluded from ACCA for that reason. For example, it does not comport with any conceivable congressional intent to disqualify an unenumerated crime that is most analogous to arson and presents nowhere near the risk of injury posed by arson, but presents a far greater risk of injury than burglary, which Congress has explicitly included. Thus, for what is probably the vast majority of cases, today’s opinion provides no guidance whatever, leaving the lower courts to their own devices in deciding, crime-by-crime, which conviction “involves conduct that presents a serious potential risk of physical injury to another.” It will take decades, and dozens of grants of certiorari, to allocate all the Nation’s crimes to one or the other side of this entirely reasonable and entirely indeterminate line. Compare ante, at 204 (concluding that attempted burglary poses sufficient risk), with Leocal v. Ashcroft, 543 U. S. 1 (2004) (concluding that driving under the influence of alcohol does not pose a “substantial risk that physical force against the person or property of another may be used,” 18 U.S. C. § 16(b)).

Imprecision and indeterminacy are particularly inappropriate in the application of a criminal statute. Years of prison hinge on the scope of ACCA’s residual provision, yet its boundaries are ill defined. If we are not going to deny effect to this statute as being impermissibly vague, see Part III, infra, we have the responsibility to derive from the text rules of application that will provide notice of what is covered and prevent arbitrary or discriminatory sentencing. See Kolender v. Lawson, 461 U. S. 352, 357 (1983). Offenders should be on notice that a particular course of conduct will result in a mandatory minimum prison term of 15 years. The Court prefers to keep them guessing.

II

The residual provision of clause (ii) of ACCA’s definition of violent felony — the clause that sweeps within ACCA’s ambit any crime that “otherwise involves conduct that presents a *217serious potential risk of physical injury to another” — is, to put it mildly, not a model of clarity. I do not pretend to have an all-encompassing solution that provides for crystal-clear application of the statute in all contexts. But we can do much better than today’s opinion with what Congress has given us.

A

The Eleventh Circuit properly sought to resolve this case by employing the “categorical approach” of looking only to the statutory elements of attempted burglary. See 430 F. 3d, at 1154, 1156-1157. This “generally prohibits the later court from delving into particular facts disclosed by the record of conviction, thus leaving the court normally to ‘look only to the fact of conviction and the statutory definition of the prior offense.’” Shepard v. United States, 544 U. S. 13, 17 (2005) (quoting Taylor v. United States, 495 U. S. 575, 602 (1990)). As the Court does, ante, at 202,1 would also begin with this approach.

The Government would have us provide some certainty— at least enough to decide the present ease — by holding that the attempt to commit a crime of violence should be treated the same as the completed crime. It points out that various federal laws, and many state laws, punish attempt with the same sanction as the completed crime. See Brief for United States 18-20. This would be persuasive if punishment were meted out solely on the basis of the risk of physical injury that a crime presents. It seems to me, however, that similar punishment does not necessarily imply similar risk; it more likely represents a judgment that the two crimes display a similar degree of depravity deserving of punishment or needful of deterrence. A person guilty of attempted burglary may not have placed anyone at physical risk, but he was just as willing to do so as the successful burglar. It seems to me impossible to say that equivalence of punishment suggests equivalence of imposed risk. I therefore look elsewhere for some clarification of the statutory text.

*218First to invite analysis is the word Congress placed at the forefront of the residual provision: “otherwise.” When used as an adverb (as it is in § 924(e)(2)(B)(ii), modifying the verb “involves”), “otherwise” is defined as “[i]n a different manner” or “in another way.” Webster’s New International Dictionary 1729 (2d ed. 1954). Thus, the most natural reading of the statute is that committing one of the enumerated crimes (burglary, arson, extortion, or crimes involving explosives) is one way to commit a crime “involving] conduct that presents a serious potential risk of physical injury to another”; and that other ways of committing a crime of that character similarly constitute “violent felon[ies].” In other words, the enumerated crimes are examples of what Congress had in mind under the residual provision, and the residual provision should be interpreted with those examples in mind. This commonsense principle of construction is sometimes referred to as the canon of ejusdem generis: “jW]hen a general word or phrase follows a list of specific persons or things, the general word or phrase will be interpreted to include only persons or things of the same type as those listed.” Black’s Law Dictionary 535 (7th ed. 1999) (Black’s); see, e. g., Washington State Dept. of Social and Health Servs. v. Guardianship Estate of Keffeler, 537 U. S. 371, 384-385 (2003). In this case, the application of that principle suggests that what the residual provision means by the general phrase “conduct that presents a serious potential risk of physical injury to another” is conduct that resembles, insofar as the degree of such risk is concerned, the previously enumerated crimes.1

*219In another context, I might conclude that any degree of risk that is merely similar, even if slightly less, would qualify. Obviously, such an interpretation would leave a good deal of ambiguity on the downside: How low on the risk scale can one go before the risk becomes too dissimilar from the enumerated crimes? Since the text sets forth no criterion, courts might vary dramatically in their answer. Cf. Leocal, 543 U. S. 1 (reversing the Eleventh Circuit’s determination that driving under the influence of alcohol qualifies as a crime of violence under 18 U. S. C. § 16). Where it is reasonably avoidable, such indeterminateness is unacceptable in the context of criminal sanctions. The rule of lenity, grounded in part on the need to give “ ‘fair warning’ ” of what is encompassed by a criminal statute, United States v. Bass, 404 U. S. 336, 348 (1971) (quoting McBoyle v. United States, 283 U. S. 25, 27 (1931)), demands that we give this text the more narrow reading of which it is susceptible. The requirement that the degree of risk be similar to that for the enumerated crimes means that it be no lesser than the risk posed by the least dangerous of those enumerated crimes.

B

I would turn, then, to the next logical question: Which of the four enumerated crimes — burglary, arson, extortion, or crimes involving use of explosives — poses the least “serious potential risk of physical injury to another”? The two that involve use of fire or explosives cannot possibly qualify. Thus, the question I must address is whether burglary or extortion poses a lesser risk. To do so, I must first define those crimes.

In Taylor, we defined “burglary” as used in the very provision of ACCA at issue here. We first determined that “‘burglary’ in §924(e) must have some uniform definition independent of the labels employed by the various States’ criminal codes.” 495 U. S., at 592. We considered but rejected the common-law definition, finding that “the contem*220porary understanding of ‘burglary’ has diverged a long way from its common-law roots.” Id., at 593. Ultimately, we concluded that “Congress meant by ‘burglary’ the generic sense in which the term is now used in the criminal codes of most States.” Id., at 598. To determine that sense, we looked for guidance to 2 W. LaFave & A. Scott, Substantive Criminal Law (1986), and the American Law Institute’s Model Penal Code (1980). We defined “burglary” as “any crime, regardless of its exact definition or label, having the basic elements of unlawful or unprivileged entry into, or remaining in, a building or structure, with intent to commit a crime.” Taylor, supra, at 599.

In defining “extortion” for purposes of ACCA, I would follow the same approach. “At common law, extortion was a property offense committed by a public official who took ‘any money or thing of value’ that was not due to him under the pretense that he was entitled to such property by virtue of his office.” Scheidler v. National Organization for Women, Inc., 537 U. S. 393, 402 (2003) (quoting 4 W. Blackstone, Commentaries on the Laws of England 141 (1769), and citing 3 R. Anderson, Wharton’s Criminal Law and Procedure § 1393, pp. 790-791 (1957)); see also 3 W. LaFave, Substantive Criminal Law §20.4 (2d ed. 2003). As with burglary, however, modern conceptions of extortion have gone well beyond the common-law understanding. In the Hobbs Act, for example, Congress “explicitly ‘expanded the common-law definition of extortion to include acts by private individuals.’” Scheidler, supra, at 402 (quoting Evans v. United States, 504 U. S. 255, 261 (1992)). And whereas the Hobbs Act retained the common-law requirement that something of value actually be acquired by the extortionist, Scheidler, supra, at 404-405, the majority of state statutes require only “that the defendant make a threat with intent thereby to acquire the victim’s property,” 3 LaFave, Substantive Criminal Law *221§ 20.4(a)(1), at 199 (emphasis added). Further, under most state statutes, the category of qualifying threats has expanded dramatically, to include threats to: “kill the victim in the future,” “cause economic harm,” “'bring about or continue a strike, boycott or other collective unofficial action,’ ” “unlawfully detain,” “accuse the victim of a crime,” “expose some disgraceful defect or secret of the victim which, when known, would subject him to public ridicule or disgrace,” and “impair one’s credit or business repute.” Id., § 20.4(a)(4), at 200, 201.

The Model Penal Code’s definition of “Theft by Extortion” reflects this expansive modern notion of the crime:

“A person is guilty of theft [by extortion] if he purposely obtains property of another by threatening to:

“(1) inflict bodily injury on anyone or commit any other criminal offense; or

“(2) accuse anyone of a criminal offense; or

“(3) expose any secret tending to subject any person to hatred, contempt or ridicule, or to impair his credit or business repute; or

“(4) take or withhold action as an official, or cause an official to take or withhold action; or

“(5) bring about or continue a strike, boycott or other collective unofficial action, if the property is not demanded or received for the benefit of the group in whose interest the actor purports to act; or

“(6) testify or provide information or withhold testimony or information with respect to another’s legal claim or defense; or

“(7) inflict any other harm which would not benefit the actor.” § 223.4, at 201.

Other federal statutes, including the Hobbs Act, 18 U. S. C. § 1951, the Travel Act, § 1952 (2000 ed. and Supp. IV), and the Racketeer Influenced and Corrupt Organizations Act *222(RICO), § 1961 et seq., use a similarly broad conception of extortion. See United States v. Nardello, 393 U. S. 286 (1969) (Travel Act); Scheidler, supra (Hobbs Act and RICO).2

The word “extortion” in ACCA’s definition of “violent felony” cannot, however, incorporate the full panoply of threats that would qualify under the Model Penal Code, many of which are inherently nonviolent. I arrive at this conclusion for two reasons: First, another canon of statutory construction, noscitur a sociis, which counsels that “the meaning of an unclear word or phrase should be determined by the words immediately surrounding it.” Black’s 1084; see Keffeler, 537 U. S., at 384-385. Of course noscitur a sociis is just an erudite (or some would say antiquated) way of saying what common sense tells us to be true: “[A] word is known by the company it keeps,” Jarecki v. G. D. Searle & Co., 367 U. S. 303, 307 (1961) — that is to say, which of various possible meanings a word should be given must be determined in a manner that makes it “fit” with the words with which it is closely associated. The words immediately surrounding “extortion” in § 924(e)(2)(B)(ii) are “burglary,” “arson,” and crimes “involv[ing] use of explosives.” The Model Penal Code’s sweeping definition of extortion would sit uncomfortably indeed amidst this list of crimes which, as the “otherwise” residual provision makes plain, are characterized by their potential for violence and their risk of physical harm to others. ACCA’s usage of “extortion” differs from the con*223text in which the word appears in the Travel Act, where it is one of a list of crimes “often used by organized crime to collect. . . revenue,” Nardello, supra, at 291, n. 8, including bribery. And it differs from the context in which “extortion” appears in RICO, where it is part of a laundry list of nearly every federal crime under the sun. See 18 U. S. C. § 1961(1) (2000 ed., Supp. IV).3

What is suggested by the canon is reinforced by the fact that both the original common-law notion of extortion, and the full expanse of the modern definition, include crimes so inherently unlikely to cause physical harm that it would set the bar of the residual provision at a level that could embrace virtually any offense — making the limitation to “serious potential risk of physical injury to another” utterly incomprehensible.4 See Part III, infra. I therefore assume that ex*224tortion under ACCA is: the obtaining of something of value from another, with his consent, induced by the wrongful use or threatened use of force against the person or property of another. Cf. Leocal, 543 U. S., at 13 (discussing the relationship between the “use of force against the person or property of another” and “crime[s] of violence under 18 U. S. C. § 16”).

One final consideration is worthy of mention. I must make sure that my restricted definition of generic extortion does not render the inclusion of extortion in § 924(e)(2)(B)(ii) superfluous in light of § 924(e)(2)(B)(i). “It is our duty to give effect, if possible, to every clause and word of a statute.” Duncan v. Walker, 533 U. S. 167, 174 (2001) (internal quotation marks omitted). Clause (i) already includes in ACCA’s definition of “violent felony” any crime that “has as an element the use, attempted use, or threatened use of physical force against the person of another.” My narrow definition of extortion passes the surplusage test only if it includes crimes that would not be covered by this provision. That is not a problem, since my definition includes the use or threatened use of force against property, whereas clause (i) is limited to force against the person. Thus, the obtaining of someone else’s money by threatening to wreck his place of business would fit within clause (ii) but not within clause (i).

Having defined burglary and extortion, I return to the question that launched this investigation in the first place: Which of the two poses the least “serious potential risk of physical injury to another”? Recall the definitions: burglary is the “unlawful or unprivileged entry into, or remaining in, a building or structure, with intent to commit a crime”; extortion is “the obtaining of something of value *225from another, with his consent, induced by the wrongful use or threatened use of force against the person or property of another.” Every victim of extortion is the object of a threat, to his person or his property; if he ignores that threat, or resists it by seeking to protect his property, he may be harmed. Burglary, on the other hand, involves only the possibility that a confrontation will take place while the crime is underway; the risk of physical harm can become a reality only if the property owner happens to be present, a situation which the burglar ordinarily seeks to avoid. The extortionist, moreover, has already expressed his willingness to commit a violent act; the burglar may be prepared to flee at the first sign of human presence. I think it obvious that burglary is less inherently risky than extortion, and thus the least inherently risky of the four crimes enumerated in § 924(e)(2)(B)(ii).

C

Having concluded in Part II-A that a crime may qualify under ACCA’s violent felony residual provision only if it poses at least as much risk of physical injury to another as the least risky of the enumerated crimes; and in Part II-B that the least risky of the enumerated crimes is burglary; I am finally able to turn to the ultimate question posed by this ease: Does attempted burglary categorically qualify as a violent felony under ACCA’s residual provision? Or as my analysis has recast that question, does attempted burglary categorically involve conduct that poses at least as much risk of physical injury to another as completed burglary? Contrary to what the Court says, ante, at 203-207, the answer must be no.

In Taylor, we discussed the risks posed by the conduct involved in a completed burglary. We found it significant that a burglary involves “invasion of victims’ homes or workplaces,” 495 U. S., at 581 (internal quotation marks and brackets omitted), and we dwelled on such an invasion’s “inherent potential for harm to persons,” id., at 588. In com*226paring attempted burglary to completed burglary, the Court focuses almost exclusively on “the possibility of a face-to-face confrontation between the burglar and a third party. ” Ante, at 203. But it ignores numerous other factors that make a completed burglary far more dangerous than a failed one: the closer proximity between burglar and victim where a confrontation takes place inside the confined space of the victim’s home; the greater likelihood of the victim’s initiating violence inside his home to protect his family and property; the greater likelihood that any confrontation inside the home will be between the burglar and the occupant of the home, rather than the police. The so-called “confrontation” the Court envisions between a would-be burglar and a third party while the burglar is still outside the home is likely to consist of nothing more than the occupant’s yelling “Who’s there?” from his window, and the burglar’s running away. It is simply not the case, as the Court apparently believes, that would-be home entries are often reduced to attempted home entries by physical confrontation between homeowner and criminal while the latter is still outside the house. (One must envision a householder throwing open his front door, shotgun in hand, just as the would-be burglar is trying to pick the lock.)

As we have previously stated, it is “[t]he fact that an offender enters a building to commit a crime [that] creates the possibility of a violent confrontation between the offender and an occupant, caretaker, or some other person who comes to investigate.” Taylor, supra, at 588 (emphasis added); see also Leocal, supra, at 10 (“[B]urglary, by its nature, involves a substantial risk that the burglar will use force against a victim in completing the crime” (emphasis added)). By definition, a perpetrator who has been convicted only of attempted burglary has failed to make it inside the home or workplace. (Indeed, a criminal convicted only of attempted burglary almost certainly injured no one; otherwise, he *227would have been convicted of something far more serious, such as assault or murder.) Thus, the full extent of the risk that burglary poses — the entry into the home that makes burglary such a threat to the physical safety of its victim— is necessarily absent in attempted burglary, however “attempt” is defined.

Because attempted burglary categorically poses a less “serious potential risk of physical injury to another” than burglary, the least risky of ACCA’s enumerated crimes, I would hold that it cannot be a predicate “violent felony” for purposes of ACCA’s mandatory minimum sentencing enhancement, § 924(e) (2000 ed. and Supp. IV), regardless of how close a State’s attempt statute requires the perpetrator come to completing the underlying offense.5

D

The Court observes, with undoubted accuracy, that my approach is not perfect. It leaves it to courts to decide, “[without hard statistics” to guide them, ante, at 210, the degree of risk of physical injury posed by various crimes. But this is an imponderable that cannot be avoided when dealing with a statute that turns upon “a serious potential risk of physical injury to another.” It inheres in the Court’s puny solution as well (how does the Court know that attempted burglary poses the same risk as burglary?). What this dissent must establish is not that my solution is perfect, but that it is substantially better than what the Court proposes. And there is little doubt of that. For in addition to *228leaving up in the air for judicial determination how much risk of physical injury each crime presents, the Court’s uninformative opinion leaves open, to be guessed at by lower courts and by those subjected to this law: (1) whether the degree of risk covered by the residual provision is limited by the degrees of risk presented by the enumerated crimes;6 (2) if so, whether extortion is to be given its broadest meaning, which would embrace crimes with virtually no risk of physical injury; and most importantly (3) where in the world to set the minimum risk of physical injury that will qualify. This indeed leaves the lower courts and those subject to this law to sail upon a virtual sea of doubt. The only thing the Court decides (and that, in my view, erroneously) is that attempted burglary poses the same risk of physical injury as burglary, and hence is covered without the need to address these other bothersome questions (how wonderfully convenient!).

It is only the Court’s decision-averse solution that enables it to accuse me of “unnecessarily deciding] ” the meaning of extortion, ante, at 209. The Court accurately, but quite irrelevantly, asserts the following:

“[E]ven if an unenumerated offense could not qualify without presenting at least as much risk as the least risky of the enumerated offenses, it would not be necessary to identify the least risky of those offenses in order to decide this case. Rather, it would be sufficient to establish simply that the unenumerated offense pre*229sented at least as much risk as one of the enumerated offenses.” Ante, at 209-210 (emphasis added).

That is true enough, and I would be properly criticized for reaching an unnecessary question if, like the Court, I found attempted burglary to be as risky as burglary. Since I do not, however, it is unavoidable that I determine the meaning of extortion, in order to decide whether attempted burglary is less risky than that. The Court’s criticism amounts to nothing more than a procedural quibble: Instead of deciding, as I have, (1) that arson and the use of explosives are the most risky of the enumerated crimes; (2) that as between burglary and extortion, burglary is the less risky (a determination requiring me to decide the meaning of extortion); and finally (3) that attempted burglary is less risky than burglary, I should have decided (1) that attempted burglary is less risky than arson, the use of explosives, and burglary; and only then (2) that attempted burglary is less risky than extortion (a determination requiring me to decide the meaning of extortion). Perhaps so, but it is surely a distinction without a real-world difference. Under either approach, determining the meaning of extortion is unquestionably necessary.

Ill

Congress passed ACCA to enhance punishment for gun-wielding offenders who have, inter alia, previously committed crimes that pose a “serious potential risk of physical injury to another.” Congress provided examples of crimes that meet this eminently reasonable but entirely abstract condition. Unfortunately, however, the four examples have little in common, most especially with respect to the level of risk of physical injury they pose. Such shoddy draftsmanship puts courts to a difficult choice: They can (1) apply the ACCA enhancement to virtually all predicate offenses, see n. 4, supra; (2) apply it case by case in its pristine abstrae*230tion, finding it applicable whenever the particular sentencing judge (or the particular reviewing panel) believes there is a “serious potential risk of physical injury to another” (whatever that means); (3) try to figure out a coherent way of interpreting the statute so that it applies in a relatively predictable and administrable fashion to a smaller subset of crimes; or (4) recognize the statute for the drafting failure it is and hold it void for vagueness, see Kolender, 461 U. S., at 357; Lanzetta v. New Jersey, 306 U. S. 451, 453 (1939).

I would choose either the third option (which I have tried to implement) or the fourth, since I believe the first two to be impermissible. If Congress wanted the first — subjecting all repeat offenders to a 15-year mandatory minimum prison term — it could very easily have crafted a statute which said that. ACCA, with its tedious definition of “violent felony,” was obviously not meant to have such an effect. The second option (the one chosen by the Court today) — essentially leaving it to the courts to apply the vague language in a manner that is ex ante (if not at the end of the day) highly unpredictable — violates, in my view, the constitutional prohibition against vague criminal laws.7 Congress has simply abdicated its responsibility when it passes a criminal statute insusceptible of an interpretation that enables principled, predictable application; and this Court has abdicated its responsibility when it allows that. Today’s opinion permits *231an unintelligible criminal statute to survive uncorrected, unguided, and unexplained. I respectfully dissent.

Justice Thomas,

dissenting.

For the reasons set forth in my opinion concurring in part and concurring in the judgment in Shepard v. United Suites, 544 U. S. 13, 27 (2005), I believe that “[t]he constitutional infirmity of § 924(e)(1) as applied to [James] makes today’s decision an unnecessary exercise.” Apprendi v. New Jersey, 530 U. S. 466 (2000), and its progeny prohibit judges from “mak[ing] a finding that raises [a defendant’s] sentence beyond the sentence that could have lawfully been imposed by reference to facts found by the jury or admitted by the defendant.” United States v. Booker, 543 U. S. 220, 317-318 (2005) (Thomas, J., dissenting in part). Yet that is precisely what the Armed Career Criminal Act, 18 U. S. C. § 924(e) (2000 ed. and Supp. IV), permits in this case.

Petitioner Alphonso James pleaded guilty to being a felon in possession of a firearm, in violation of § 922(g)(1) (2000 ed.), which exposed him to a maximum sentence of 10 years under § 924(a)(2). Section 924(e)(1) (2000 ed., Supp. IV), however, mandated a minimum 15-year sentence if James had three prior convictions for “a violent felony or a serious drug offense.” James admitted he had been convicted of three prior felonies, but he argued that one of those felonies — his conviction for attempted burglary of a dwelling, in violation of Fla. Stat. §§ 810.02 and 777.04 (2006) — was not a “violent felony” for purposes of 18 U. S. C. § 924(e)(1) (2000 ed., Supp. IV). The District Court resolved this disputed fact in favor of the Government and increased James’ sentence accordingly. Relying on the scheme we initially created in Taylor v. United States, 495 U. S. 575 (1990), the Court of Appeals affirmed.

Section 924(e)(1), in conjunction with Taylor, Shepard, and now today’s decision, “explain[s] to lower courts how to conduct factfinding that is, according to the logic of this Court’s *232intervening precedents, unconstitutional in this very case.” Shepard, supra, at 27 (Thomas, J., concurring in part and concurring in judgment). For that reason, I respectfully dissent.