Larry BEGAY, Petitioner,
Supreme Court of United States.
Argued January 15, 2008.Decided April 16, 2008.
*1583 Margaret A. Katze, Albuquerque, NM, for petitioner.
Leondra R. Kruger, for respondent.
Stephen P. McCue, Federal Public Defender, Margaret A. Katze, Counsel of Record, Assistant Federal Public Defender, Charles McCormack, Research and Writing Specialist, Office of the Federal Public Defender, Albuquerque, NM, for petitioner.
Paul D. Clement, Solicitor General, Counsel of Record, Alice S. Fisher, Assistant Attorney General, Michael R. Dreeben, Deputy Solicitor General, Leondra R. Kruger, Assistant to the Solicitor General, Richard A. Friedman, Attorney, Department of Justice, Washington, D.C., for respondent.
Justice BREYER delivered the opinion of the Court.
The Armed Career Criminal Act imposes a special mandatory 15-year prison term upon felons who unlawfully possess a firearm and who also have three or more previous convictions for committing certain drug crimes or “violent felon[ies].” 18 U.S.C. ? 924(e)(1) (2000 ed., Supp. V). The question in this case is whether driving under the influence of alcohol is a “violent felony” as the Act defines it. We conclude that it is not.
Federal law prohibits a previously convicted felon from possessing a firearm. ? 922(g)(1) (2000 ed.). A related provision provides for a prison term of up to 10 years for an ordinary offender. ? 924(a)(2). The Armed Career Criminal Act imposes a more stringent 15-year mandatory minimum sentence on an offender who has three prior convictions “for a violent felony or a serious drug offense.” ? 924(e)(1) (2000 ed., Supp. V).
The Act 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) (2000 ed.).
We here consider whether driving under the influence of alcohol (DUI), as set forth in New Mexico’s criminal statutes, falls within the scope of the second clause.
The relevant background circumstances include the following: In September 2004, New Mexico police officers received a report that Larry Begay, the petitioner here, had threatened his sister and aunt with a rifle. The police arrested him. Begay subsequently conceded he was a felon and *1584 pleaded guilty to a federal charge of unlawful possession of a firearm in violation of ? 922(g)(1). Begay’s presentence report said that he had been convicted a dozen times for DUI, which under New Mexico’s law, becomes a felony (punishable by a prison term of more than one year) the fourth (or subsequent) time an individual commits it. See N.M. Stat. Ann. ?? 66-8-102(G) to (J) (Supp.2007). The sentencing judge consequently found that Begay had at least three prior convictions for a crime “punishable by imprisonment for a term exceeding one year.” 377 F. Supp. 2d 1141, 1143 (NM 2005). The judge also concluded that Begay’s “three felony DUI convictions involve conduct that presents a serious potential risk of physical injury to another.” Id., at 1145. The judge consequently concluded that Begay had three or more prior convictions for a “violent felony” and should receive a sentence that reflected a mandatory minimum prison term of 15 years. Ibid.
Begay, claiming that DUI is not a “violent felony” within the terms of the statute, appealed. The Court of Appeals panel by a vote of 2 to 1 rejected that claim. 470 F.3d 964 (C.A.10 2006). Begay sought certiorari, and we agreed to decide the question.
New Mexico’s DUI statute makes it a crime (and a felony after three earlier convictions) to “drive a vehicle within [the] state” if the driver “is under the influence of intoxicating liquor” (or has an alcohol concentration of .08 or more in his blood or breath within three hours of having driven the vehicle resulting from “alcohol consumed before or while driving the vehicle”). ?? 66-8-102(A), (C). In determining whether this crime is a violent felony, we consider the offense generically, that is to say, we examine it in terms of how the law defines the offense and not in terms of how an individual offender might have committed it on a particular occasion. See Taylor v. United States, 495 U.S. 575, 602, 110 S. Ct. 2143, 109 L. Ed. 2d 607 (1990) (adopting this “categorical approach”); see also James v. United States, 550 U.S. ___, ___, 127 S. Ct. 1586, 1597, 167 L. Ed. 2d 532 (2007) (attempted burglary is a violent felony even if, on some occasions, it can be committed in a way that poses no serious risk of physical harm).
We also take as a given that DUI does not fall within the scope of the Act’s clause (i) “violent felony” definition. DUI, as New Mexico defines it, nowhere “has 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).
Finally, we assume that the lower courts were right in concluding that DUI involves conduct that “presents a serious potential risk of physical injury to another.” ? 924(e)(2)(B)(ii). Drunk driving is an extremely dangerous crime. In the United States in 2006, alcohol-related motor vehicle crashes claimed the lives of more than 17,000 individuals and harmed untold amounts of property. National Highway Traffic Safety Admin., Traffic Safety Facts, 2006 Traffic Safety Annual Assessment?Alcohol-Related Fatalities 1 (No. 810821, Aug. 2007), http://www-nrd.nhtsa. dot.gov/Pubs/810821.PDF (as visited Apr. 11, 2008, and available in Clerk of Court’s case file). Even so, we find that DUI falls outside the scope of clause (ii). It is simply too unlike the provision’s listed examples for us to believe that Congress intended the provision to cover it.
In our view, the provision’s listed examples?burglary, arson, extortion, or *1585 crimes involving the use of explosives? illustrate the kinds of crimes that fall within the statute’s scope. Their presence indicates that the statute covers only similar crimes, rather than every crime that “presents a serious potential risk of physical injury to another.” ? 924(e)(2)(B)(ii). If Congress meant the latter, i.e., if it meant the statute to be all-encompassing, it is hard to see why it would have needed to include the examples at all. Without them, clause (ii) would cover all crimes that present a “serious potential risk of physical injury.” Ibid. Additionally, if Congress meant clause (ii) to include all risky crimes, why would it have included clause (i)? A crime which has as an element the “use, attempted use, or threatened use of physical force” against the person (as clause (i) specifies) is likely to create “a serious potential risk of physical injury” and would seem to fall within the scope of clause (ii).
Of course, Congress might have included the examples solely for quantitative purposes. Congress might have intended them to demonstrate no more than the degree of risk sufficient to bring a crime within the statute’s scope. But were that the case, Congress would have likely chosen examples that better illustrated the “degree of risk” it had in mind. Our recent case, James v. United States?where we considered only matters of degree, i.e., whether the amount of risk posed by attempted burglary was comparable to the amount of risk posed by the example crime of burglary?illustrates the difficulty of interpreting the examples in this respect. Compare 550 U.S., at ___ _ ___, 127 S.Ct., at 1594-1597, with id., at ___, ___ _ ___, ___, 127 S.Ct., at 1601, 1603-1604, 1609 (SCALIA, J., dissenting). Indeed, the examples are so far from clear in respect to the degree of risk each poses that it is difficult to accept clarification in respect to degree of risk as Congress’ only reason for including them. See id., at 1598-99 (“Congress provided examples [that] … have little in common, most especially with respect to the level of risk of physical injury that they pose”).
These considerations taken together convince us that, “`to give effect … to every clause and word'” of this statute, we should read the examples as limiting the crimes that clause (ii) covers to crimes that are roughly similar, in kind as well as in degree of risk posed, to the examples themselves. Duncan v. Walker, 533 U.S. 167, 174, 121 S. Ct. 2120, 150 L. Ed. 2d 251 (2001) (quoting United States v. Menasche, 348 U.S. 528, 538-539, 75 S. Ct. 513, 99 L. Ed. 615 (1955); some internal quotation marks omitted); see also Leocal v. Ashcroft, 543 U.S. 1, 12, 125 S. Ct. 377, 160 L. Ed. 2d 271 (2004) (describing the need to interpret a statute in a way that gives meaning to each word).
The concurrence complains that our interpretive approach is insufficiently specific. See post, at 1589-1590 (SCALIA, J., concurring in judgment). But the concurrence’s own approach demands a crime-by-crime analysis, uses a standard of measurement (comparative degree of risk) that even the concurrence admits is often “unclear,” post, at 1590, requires the concurrence to turn here to the still less clear “rule of lenity,” post, at 1591, and, as we explain, is less likely to reflect Congress’ intent. See, e.g., post, at 1590-1591 (recognizing inability to measure quantitative seriousness of risks associated with DUI).
The statute’s history offers further support for our conclusion that the examples in clause (ii) limit the scope of the clause to crimes that are similar to the examples themselves. Prior to the enactment of the current language, the Act applied its enhanced sentence to offenders with “three *1586 previous convictions for robbery or burglary.” Taylor, 495 U.S., at 581, 110 S. Ct. 2143 (internal quotation marks omitted). Congress sought to expand that definition to include both crimes against the person (clause (i)) and certain physically risky crimes against property (clause (ii)). See H.R.Rep. No. 99-849, p. 3 (1986) (hereinafter H.R.Rep.). When doing so, Congress rejected a broad proposal that would have covered every offense that involved a substantial risk of the use of “`physical force against the person or property of another.'” Taylor, supra, at 583, 110 S. Ct. 2143 (quoting S. 2312, 99th Cong., 2d Sess. (1986); H.R. 4639, 99th Cong., 2d Sess. (1986)). Instead, it added the present examples. And in the relevant House Report, it described clause (ii) as including “State and Federal felonies 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.” H.R. Rep., at 5 (emphasis added).
Of course, the statute places the word “otherwise,” just after the examples, so that the provision covers a felony that is one of the example crimes “or otherwise involves conduct that presents a serious potential risk of physical injury.” ? 924(e)(2)(B)(ii) (emphasis added). But we cannot agree with the Government that the word “otherwise” is sufficient to demonstrate that the examples do not limit the scope of the clause. That is because the word “otherwise” can (we do not say must, cf. post, at 1589-1590 (SCALIA, J., concurring in judgment)) refer to a crime that is similar to the listed examples in some respects but different in others?similar say in respect to the degree of risk it produces, but different in respect to the “way or manner” in which it produces that risk. Webster’s Third New International Dictionary 1598 (1961) (defining “otherwise” to mean “in a different way or manner”).
In our view, DUI differs from the example crimes?burglary, arson, extortion, and crimes involving the use of explosives?in at least one pertinent, and important, respect. The listed crimes all typically involve purposeful, “violent,” and “aggressive” conduct. 470 F.3d, at 980 (McConnell, J., dissenting in part); see, e.g., Taylor, supra, at 598, 110 S. Ct. 2143 (“burglary” is an unlawful or unprivileged entry into a building or other structure with “intent to commit a crime”); ALI Model Penal Code ? 220.1(1) (1985) (“arson” is causing a fire or explosion with the purpose of,” e.g., “destroying a building… of another” or “damaging any property … to collect insurance”); id., ? 223.4 (extortion is “purposely” obtaining property of another through threat of, e.g., inflicting “bodily injury”); Leocal, supra, at 9, 125 S. Ct. 377 (the word “`use’ … most naturally suggests a higher degree of intent than negligent or merely accidental conduct” which fact helps bring it outside the scope of the statutory term “crime of violence”). That conduct is such that it makes more likely that an offender, later possessing a gun, will use that gun deliberately to harm a victim. Crimes committed in such a purposeful, violent, and aggressive manner are “potentially more dangerous when firearms are involved.” 470 F.3d, at 980 (McConnell, J., dissenting in part). And such crimes are “characteristic of the armed career criminal, the eponym of the statute.” Ibid.
By way of contrast, statutes that forbid driving under the influence, such as the statute before us, typically do not insist on purposeful, violent, and aggressive conduct; rather, they are, or are most nearly comparable to, crimes that impose strict liability, criminalizing conduct in respect to which the offender need not have had any *1587 criminal intent at all. The Government argues that “the knowing nature of the conduct that produces intoxication combined with the inherent recklessness of the ensuing conduct more than suffices” to create an element of intent. Brief for United States 35. And we agree with the Government that a drunk driver may very well drink on purpose. But this Court has said that, unlike the example crimes, the conduct for which the drunk driver is convicted (driving under the influence) need not be purposeful or deliberate. See Leocal, 543 U.S., at 11, 125 S. Ct. 377 (a DUI offense involves “accidental or negligent conduct”); see also 470 F.3d, at 980 (McConnell, J., dissenting in part) (“[D]runk driving is a crime of negligence or recklessness, rather than violence or aggression”).
When viewed in terms of the Act’s basic purposes, this distinction matters considerably. As suggested by its title, the Armed Career Criminal Act focuses upon the special danger created when a particular type of offender?a violent criminal or drug trafficker?possesses a gun. See Taylor, supra, at 587-588, 110 S. Ct. 2143; 470 F.3d, at 981, n. 3 (McConnell, J., dissenting in part) (“[T]he title [of the Act] was not merely decorative”). In order to determine which offenders fall into this category, the Act looks to past crimes. This is because an offender’s criminal history is relevant to the question whether he is a career criminal, or, more precisely, to the kind or degree of danger the offender would pose were he to possess a gun.
In this respect?namely, a prior crime’s relevance to the possibility of future danger with a gun?crimes involving intentional or purposeful conduct (as in burglary and arson) are different than DUI, a strict liability crime. In both instances, the offender’s prior crimes reveal a degree of callousness toward risk, but in the former instance they also show an increased likelihood that the offender is the kind of person who might deliberately point the gun and pull the trigger. We have no reason to believe that Congress intended a 15-year mandatory prison term where that increased likelihood does not exist.
Were we to read the statute without this distinction, its 15-year mandatory minimum sentence would apply to a host of crimes which, though dangerous, are not typically committed by those whom one normally labels “armed career criminals.” See, e.g., Ark.Code Ann. ? 8-4-103(a)(2)(A)(ii) (2007) (reckless polluters); 33 U.S.C. ? 1319(c)(1) (individuals who negligently introduce pollutants into the sewer system); 18 U.S.C. ? 1365(a) (individuals who recklessly tamper with consumer products); ? 1115 (seamen whose inattention to duty causes serious accidents). We have no reason to believe that Congress intended to bring within the statute’s scope these kinds of crimes, far removed as they are from the deliberate kind of behavior associated with violent criminal use of firearms. The statute’s use of examples (and the other considerations we have mentioned) indicate the contrary.
The dissent’s approach, on the other hand, would likely include these crimes within the statutory definition of “violent felony,” along with any other crime that can be said to present “a serious potential risk of physical injury.” Post, at 1592 (opinion of ALITO, J.). And it would do so because it believes such a result is compelled by the statute’s text. See ibid. But the dissent’s explanation does not account for a key feature of that text? namely, the four example crimes intended to illustrate what kind of “violent felony” the statute covers. The dissent at most believes that these examples are relevant only to define the “requisite” serious risk associated with a “crime of violence.” *1588 Post, at 1595. But the dissent does not explain what it means by “requisite,” nor does it describe how these various examples might help define that term in the context of this statute. If they were in fact helpful on that score, we might expect more predictable results from a purely risk-based approach. Compare post, at 1588, 1591-1592 (SCALIA, J., concurring in judgment), with post, at 1592-1594 (dissenting opinion). Thus, the dissent’s reliance on these examples for a function they appear incapable of performing reads them out of the statute and, in so doing, fails to effectuate Congress’ purpose to punish only a particular subset of offender, namely career criminals.
The distinction we make does not minimize the seriousness of the risks attached to driving under the influence. Nor does our argument deny that an individual with a criminal history of DUI might later pull the trigger of a gun. (Indeed, we may have such an instance before us. 470 F.3d, at 965.) Rather, we hold only that, for purposes of the particular statutory provision before us, a prior record of DUI, a strict liability crime, differs from a prior record of violent and aggressive crimes committed intentionally such as arson, burglary, extortion, or crimes involving the use of explosives. The latter are associated with a likelihood of future violent, aggressive, and purposeful “armed career criminal” behavior in a way that the former are not.
We consequently conclude that New Mexico’s crime of “driving under the influence” falls outside the scope of the Armed Career Criminal Act’s clause (ii) “violent felony” definition. And we reverse the judgment of the Court of Appeals in relevant part and remand the case for proceedings consistent with this opinion.
It is so ordered.
Justice SCALIA, concurring in the judgment.
The statute in this case defines “violent felony” in part as “any crime punishable by imprisonment for a term exceeding one year … that … is burglary, arson, or extortion, involves the use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.” 18 U.S.C. ? 924(e)(2)(B)(ii). Contrary to the Court, I conclude that the residual clause unambiguously encompasses all crimes that present a serious risk of injury to another. But because I cannot say that drunk driving clearly poses such a risk (within the meaning of the statute), the rule of lenity brings me to concur in the judgment of the Court.
Last Term, in James v. United States, 550 U.S. ___, 127 S. Ct. 1586 (2007), the Court held that attempted burglary qualifies as a violent felony under ? 924(e). It concluded that to determine whether a predicate crime falls under the residual clause, a court should first identify the enumerated crime to which the predicate crime is most closely analogous and then decide whether the risk posed by the predicate crime is roughly equivalent to the risk posed by the enumerated crime. Because burglary was the enumerated crime most closely analogous to attempted burglary, and attempted burglary in the Court’s judgment posed roughly the same risk of physical injury as burglary, attempted burglary qualified as a “violent felony” under ? 924(e). See id., at ___, 127 S.Ct., at 1597.
Unfortunately, the Court’s approach in deciding that case provided no guidance for deciding future cases that involve predicate crimes other than attempted burglary, particularly those for which there are no clear analogs among the enumerated *1589 crimes. Pointing out that problem in dissent, I anticipated this very case: “Is, for example, driving under the influence of alcohol more analogous to burglary, arson, extortion, or a crime involving use of explosives?” Id., at ___, 127 S.Ct., at 1601.
My dissent set out a different approach to the statute. In my view, the best way to interpret ? 924(e) is first to determine which of the enumerated offenses poses the least serious risk of physical injury, and then to set that level of risk as the “serious potential risk” required by the statute. Crimes that pose at least that serious a risk of injury are encompassed by the residual clause; crimes that do not are excluded. In my judgment, burglary was the least risky crime among the enumerated offenses, and I therefore concluded that attempted burglary, which is less risky than burglary, is not covered by the residual clause.
The Court held otherwise in James, and since this is a statutory case that holding has a strong claim to stare decisis. But the concomitant of the sad fact that the theory of James has very limited application is the happy fact that its stare decisis effect is very limited as well. It must be followed, I presume, for unenumerated crimes that are analogous to enumerated crimes (e.g., attempted arson). It provides no answer, and suggests no approach to an answer, where, as here, the predicate crime has no analog among the enumerated crimes. For such cases I would therefore adhere to the principles I set forth in my James dissent.
Today the Court devises a different way to give concrete meaning to the residual clause. Confronted with a predicate crime that has no obvious analog among the enumerated offenses, the Court engrafts a requirement onto the residual clause that a predicate crime involve “purposeful, `violent,’ and `aggressive’ conduct.” Ante, at 1586. By doing so, it excludes a slew of crimes from the scope of the residual clause, including (not by happenstance) the crime at issue here, drunk driving. Like James, this latest made-for-the-case improvisation does not (as my resolution does) provide a complete framework that will embrace all future cases. There are still many crimes that are not analogous to the enumerated crimes (so that their status cannot be resolved by James) but do involve “purposeful, `violent,’ and `aggressive’ conduct” (so that their status cannot be resolved by today’s deus ex machina). Presumably some third (and perhaps fourth and fifth) gimmick will be devised to resolve those cases as they arise, leaving our brethren on the district courts and courts of appeals much room for enjoyable speculation.
But quite apart from its regrettable continuation of a piecemeal, suspenseful, Scrabble-like approach to the interpretation of this statute, the problem with the Court’s holding today is that it is not remotely faithful to the statute that Congress wrote. There is simply no basis (other than the necessity of resolving the present case) for holding that the enumerated and unenumerated crimes must be similar in respects other than the degree of risk that they pose.
The Court is correct that the clause “otherwise involves conduct that presents a serious potential risk of physical injury to another” signifies a similarity between the enumerated and unenumerated crimes. It is not, however, any old similarity, such as (to take a random example) “purposeful, `violent,’ and `aggressive’ conduct.” Rather, it is the particular similarity specified after the “otherwise”?i.e., that they all pose a serious potential risk of physical injury to another. They need not be similar *1590 in any other way. As the Court correctly notes, the word “otherwise” in this context means “`in a different way or manner.’ ” Ante, at 1587; see also James, 550 U.S., at ___, 127 S.Ct., at 1602 (SCALIA, J., dissenting); Webster’s New International Dictionary 1729 (2d ed.1957) (“in another way or in other ways”). Therefore, by using the word “otherwise” the writer draws a substantive connection between two sets only on one specific dimension?i.e., whatever follows “otherwise.” What that means here is that “committing one of the enumerated crimes … is one way to commit a crime `involv[ing] 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].'” James, supra, at ___, 127 S.Ct., at 1603 (SCALIA, J., dissenting).
The Court rejects this seemingly straightforward statutory analysis, reading the residual clause to mean that the unenumerated offenses must be similar to the enumerated offenses not only in the degree of risk they pose, but also “in kind,” despite the fact that “otherwise” means that the common element of risk must be presented “`in a different way or manner.'” Ante, at 1585-1586 (emphasis added). The Court’s explanation for this interpretation seems to be that the enumerated crimes are “so far from clear in respect to the degree of risk each poses that it is difficult to accept clarification in respect to degree of risk as Congress’s only reason for including them.” Ante, at 1585. While I certainly agree that the degree of risk associated with the enumerated crimes is unclear, I find it unthinkable that the solution to that problem is to write a different statute. The phrase “otherwise involves conduct that presents a serious potential risk of physical injury to another” limits inclusion in the statute only by a crime’s degree of risk. See James, supra, at ___, 127 S.Ct., at 1603 (SCALIA, J., dissenting). The use of the adjective “serious” seems to me to signify a purely quantitative measure of risk. If both an intentional and a negligent crime pose a 50% risk of death, could one be characterized as involving a “serious risk” and the other not? Surely not.
The Court supports its argument with that ever-ready refuge from the hardships of statutory text, the (judicially) perceived statutory purpose. According to the Court, because the Armed Career Criminal Act is concerned with “the special danger created when a particular type of offender?a violent criminal or drug trafficker?possesses a gun,” the statutory purpose favors applying ? 924(e)’s enhanced penalty only to those criminals “who might deliberately point the gun and pull the trigger.” Ante, at 1587. I cannot possibly infer that purpose from the statute. For all I know, the statute was meant to punish those who are indifferent to human life, or who are undeterred by the criminal penalties attached to the commission of other crimes (after all, the statute enhances penalties for drug traffickers, see ? 924(e)(2)(A)). While the Court’s asserted purpose would surely be a reasonable one, it has no more grounding in the statutory text than do these other possibilities. And what is more, the Court’s posited purpose is positively contradicted by the fact that one of the enumerated crimes?the unlawful use of explosives?may involve merely negligent or reckless conduct. See ALI, Model Penal Code ? 220.2(2) (1985) (“A person is guilty of a misdemeanor if he recklessly creates a risk of catastrophe in the employment of fire, explosives or other dangerous means”); id., ? 220.3 (“A person is guilty of criminal mischief if he … damages tangible property of another purposely, recklessly, or by negligence in the *1591 employment of fire, explosives, or other dangerous means”).
The Court says that an interpretation of the residual clause that includes all crimes posing a serious risk of injury would render superfluous ? 924(e)(2)(B)(i), which provides that a “violent felony” is any crime that “has as an element the use, attempted use, or threatened use of physical force against the person” of another. Ante, at 1584 (internal quotation marks omitted). But the canon against surplusage has substantially less force when it comes to interpreting a broad residual clause like the one at issue here. Though the second clause renders the first superfluous, it would raise no eyebrows to refer to “crimes that entail the use of force and crimes that, while not entailing the use of force, nonetheless present a serious risk of injury to another person.” In any event, the canon against surplusage merely helps decide between competing permissible interpretations of an ambiguous statute; it does not sanction writing in a requirement that Congress neglected to think of. And finally, come to think of it, the Court’s solution does nothing whatever to solve the supposed surplusage problem. Crimes that include as an element “the use … of physical force against the person of another” are all embraced (and the reference to them thus rendered superfluous) by the requirement of “purposeful, `violent,’ and `aggressive’ conduct” that the Court invents.
Under my interpretation of ? 924(e), I must answer one question: Does drunk driving pose at least as serious a risk of physical injury to another as burglary? From the evidence presented by the Government, I cannot conclude so. Because of that, the rule of lenity requires that I resolve this case in favor of the defendant.
The Government cites the fact that in 2006, 17,062 persons died from alcohol-related car crashes, and that 15,121 of those deaths involved drivers with blood-alcohol concentrations of 0.08 or higher. See Brief for United States 17. Drunk driving is surely a national problem of great concern. But the fact that it kills many people each year tells us very little about whether a single act of drunk driving “involves conduct that presents a serious potential risk of physical injury to another.” It may well be that an even greater number of deaths occurs annually to pedestrians crossing the street; but that hardly means that crossing the street presents a serious potential risk of injury. Where the issue is “risk,” the annual number of injuries from an activity must be compared with the annual incidents of the activity. Otherwise drunk driving could be said to pose a more serious risk of physical harm than murder. In addition, drunk driving is a combination of two activities: (1) drinking and (2) driving. If driving alone results in injury in a certain percentage of cases, it could hardly be said that the entirety of the risk posed by drunk driving can be attributed to the combination. And finally, injuries to the drunk drivers themselves must be excluded from the calculus, because the statute counts only injuries to other persons.
Needless to say, we do not have these relevant statistics. And even if we did, we would still need to know similar statistics for burglary, which are probably even harder to come by. This does not mean that I will never be able to identify a crime that falls under the residual clause. For some crimes, the severity of the risk will be obvious. Crimes like negligent homicide, see ALI, Model Penal Code ? 210.4 (1980), conspiracy to commit a violent crime, id., ? 5.03 (1985), inciting to riot, 18 U.S.C. ?2101, and the production of *1592 chemical weapons, ? 229, certainly pose a more serious risk of physical injury to others than burglary. (By contrast, the Court’s approach eliminates from the residual clause all negligent crimes, even those that entail a 100% risk of physical injury such as negligent homicide.) But I can do no more than guess as to whether drunk driving poses a more serious risk than burglary, and I will not condemn a man to a minimum of 15 years in prison on the basis of such speculation. See Ladner v. United States, 358 U.S. 169, 178, 79 S. Ct. 209, 3 L. Ed. 2d 199 (1958). Applying the rule of lenity to a statute that demands it, I would reverse the decision of the Court of Appeals.
Justice ALITO, with whom Justice SOUTER and Justice THOMAS join, dissenting.
The statutory provision at issue in this case?the so-called “residual clause” of 18 U.S.C. ? 924(e)(2)(B)(ii)?calls out for legislative clarification, and I am sympathetic to the result produced by the Court’s attempt to craft a narrowing construction of this provision. Unfortunately, the Court’s interpretation simply cannot be reconciled with the statutory text, and I therefore respectfully dissent.
In September 2004, after a night of heavy drinking, petitioner pointed a rifle at his aunt and threatened to shoot if she did not give him money. When she replied that she did not have any money, petitioner repeatedly pulled the trigger, but the rifle was unloaded and did not fire. Petitioner then threatened his sister in a similar fashion.
At the time of this incident, petitioner was a convicted felon. He had 12 prior convictions in New Mexico for driving under the influence of alcohol (DUI). While DUI is generally a misdemeanor under New Mexico law, the offense of DUI after at least three prior DUI convictions is a felony requiring a sentence of 18 months’ imprisonment. N.M. Stat. Ann. ? 66-8-102(G) (Supp.2007).
Petitioner pleaded guilty to possession of a firearm by a convicted felon, in violation of 18 U.S.C. ? 922(g)(1). A violation of that provision generally carries a maximum term of imprisonment of 10 years, see ? 924(a)(2), but the District Court and the Court of Appeals held that petitioner was subject to a mandatory minimum sentence of 15 years because he had at least three prior convictions for the New Mexico felony of DUI after being convicted of DUI on at least three prior occasions. 377 F. Supp. 2d 1141, 1143-45 (NM 2005); 470 F.3d 964, 966-975, 977 (C.A.10 2006). The lower courts concluded that these offenses were crimes “punishable by imprisonment for a term exceeding one year” and “involve[d] conduct that present[ed] a serious potential risk of physical injury to another.” 18 U.S.C. ? 924(e)(2)(B).
The Court does not hold that the maximum term of imprisonment that petitioner faced on his felony DUI convictions was less than one year. Nor does the Court dispute that petitioner’s offenses involved “a serious potential risk of physical injury to another.” Ibid. The only remaining question, therefore, is whether the risk presented by petitioner’s qualifying DUI felony convictions was “serious,” i.e., “significant” or “important.” See, e.g., Webster’s *1593 Third New International Dictionary 2073 (hereinafter Webster’s); (2002) 15 Oxford English Dictionary 15 (def. 6(a)) (2d ed.1989) (hereinafter OED). In my view, it was.
Statistics dramatically show that driving under the influence of alcohol is very dangerous. Each year, approximately 15,000 fatal alcohol-related crashes occur, accounting for roughly 40% of all fatal crashes. Approximately a quarter million people are injured annually in alcohol-related crashes. The number of people who are killed each year by drunk drivers is far greater than the number of murders committed during any of the crimes specifically set out in the statutory provision at issue here, ? 924(e)(2)(B)(ii)?burglary, arson, extortion, and offenses involving the use of explosives.
Petitioner’s qualifying offenses, moreover, fell within the statute only because he had been convicted of DUI on at least three prior occasions. As noted, petitioner had a dozen prior DUI convictions. Persons who repeatedly drive drunk present a greatly enhanced danger that they and others will be injured as a result. In *1594 addition, it has been estimated that the ratio of DUI incidents to DUI arrests is between 250 to 1 and 2,000 to 1. Accordingly, the risk presented by a 10th, 11th, and 12th DUI conviction may be viewed as the risk created by literally thousands of drunk-driving events. That risk was surely “serious,” and therefore petitioner’s offenses fell squarely within the language of the statute.
Moreover, taking the statutory language to mean what it says would not sweep in all DUI convictions. Most DUI convictions are not punishable by a term of imprisonment of more than one year and thus fall outside the scope of the statute. Petitioner’s convictions qualified only because of his extraordinary?and, I would say, extraordinarily dangerous?record of drunk driving.
The Court holds that an offense does not fall within the residual clause unless it is “roughly similar, in kind as well as in degree of risked posed,” ante, at 1585, to the crimes specifically listed in 18 U.S.C. ? 924(e)(2)(B), i.e., burglary, extortion, arson, and crimes involving the use of explosives. These crimes, according to the Court, “all typically involve purposeful, `violent,’ and `aggressive’ conduct.” Ante, at 1586 (quoting 470 F.3d, at 980 (McConnell, J., dissenting)).
This interpretation cannot be squared with the text of the statute, which simply does not provide that an offense must be “purposeful,” “violent,” or “aggressive” in order to fall within the residual clause. Rather, after listing burglary, arson, extortion, and explosives offenses, the statute provides (in the residual clause) that an offense qualifies if it “otherwise involves conduct that presents a serious potential risk of physical injury to another.” Therefore, offenses falling within the residual clause must be similar to the named offenses in one respect only: They must, “otherwise”?which is to say, “in a different manner,” 10 OED 984 (def. B(1)); see also Webster’s 1598?”involve conduct that presents a serious potential risk of physical injury to another.” Requiring that an offense must also be “purposeful,” “violent,” or “aggressive” amounts to adding new elements to the statute, but we “ordinarily resist reading words or elements into a statute that do not appear on its face.” Bates v. United States, 522 U.S. 23, 29, 118 S. Ct. 285, 139 L. Ed. 2d 215 (1997).
Each part of this additional, judicially added requirement presents other problems as well.
Purposeful. At least one State’s DUI law requires proof of purposeful conduct. See Tam v. State, 232 Ga.App. 15, 16, 501 S.E.2d 51, 52 (1998) (requiring proof of the intent to drive). In addition, many States recognize involuntary intoxication as a defense. See 4 R. Essen & R. Erwin, Defense of Drunk Driving Cases: Criminal? Civil ? 44.04 (2007). And even in States that do not require purposefulness, I have no doubt that the overwhelming majority of DUI defendants purposefully drank before getting behind the wheel and were purposefully operating their vehicles at the time of apprehension. I suspect that many DUI statutes do not require proof of *1595 purposefulness because the element is almost always present, requiring proof of the element would introduce an unnecessary complication, and it would make no sense to preclude conviction of those defendants who were so drunk that they did not even realize that they were behind the wheel.
Violent. It is clear that 18 U.S.C. ? 924(e)(2)(B) is not limited to “violent” crimes, for if it were, it would be redundant. The prior subparagraph, ? 924(e)(2)(A), includes offenses that have as an element the use or threatened use of violence.
Aggressive. The concept of “aggressive” crimes is vague, and in any event, it is hardly apparent why DUI?not to mention the species of felony DUI recidivism that resulted in petitioner’s predicament?is not “aggressive.” Driving can certainly involve “aggressive” conduct. Indeed, some States have created the offense of “aggressive driving.” See M. Savage, M. Sundeen, & A. Teigen, Traffic Safety and Public Health: State Legislative Action 2007, Transportation Series (National Conference of State Legislatures, Dec. 2007, No. 32), p. 17, and App. J, online at http:// www.ncsl.org/print/transportation/07traffic safety.pdf. Most States have a toll-free telephone number to call to report “aggressive” driving. See Campaign Safe & Sober, Phone Numbers for Reporting Impaired, Aggressive, or Unsafe Driving, online at http://www.nhtsa.dot.gov/people/ outreach/safesobr/16qp/phone.html.
The Court defends its new statutory element on the ground that a defendant who merely engages in felony drunk driving is not likely to be “the kind of person who might deliberately point the gun and pull the trigger.” Ante, at 1587. The Court cites no empirical support for this conclusion, and its accuracy is not self-evident. Petitioner’s pattern of behavior may or may not be typical of those defendants who have enough DUI convictions to qualify under N.M. Stat. Ann. ? 66-8-102(G) and 18 U.S.C. ? 924(e)(2)(B), but the example of his behavior in this case?pointing a gun at his aunt’s head and repeatedly pulling the trigger?should surely be enough to counsel against uncritical reliance on stereotypes about “the type” of people who commit felony DUI violations.
Defendants who qualify for an enhanced sentence under ? 924(e) (2000 ed. and Supp. V) based (in whole or in part) on felony DUI convictions share at least three characteristics that are relevant for present purposes. First, they are persons who, in the judgment of Congress, cannot be trusted to use a firearm responsibly. In order to qualify for an enhanced sentence under ? 924(e), a defendant must of course be convicted of violating the felon-in-possession statute, ? 922(g) (2000 ed.). The felon-in-possession statute necessarily rests on the judgment that a person with a prior felony conviction cannot be trusted with a firearm. See Caron v. United States, 524 U.S. 308, 315, 118 S. Ct. 2007, 141 L. Ed. 2d 303 (1998) (“Congress meant to keep guns away from all offenders who, the Federal Government feared, might cause harm …”). And there is no dispute that a prior felony DUI conviction qualifies as a felony under the felon-in-possession law. If Congress thought that a person with a prior felony DUI conviction is not “the kind of person” who is likely to use a gun unlawfully, why would Congress have made it a crime for such a person to possess a gun?
Second, defendants with DUI convictions that are counted under 18 U.S.C. ? 924(e)(2)(B) are likely to have serious alcohol abuse problems. As previously mentioned, ordinary DUI convictions are generally not counted under ? 924(e) because they are not punishable by imprisonment *1596 for more than a year. Such penalties are generally reserved for persons, like petitioner, with a record of repeated DUI violations. See National Conference of State Legislatures, supra. Such individuals are very likely to have serious alcohol abuse problems and a propensity to engage in irresponsible conduct while under the influence. Alcohol use often precedes violent crimes, see, e.g., Roizen, Epidemiological Issues in Alcohol-Related Violence, in 13 Recent Developments in Alcoholism 7, 8-9 (M. Galanter ed.1997), and thus there is reason to worry about the misuse of firearms by defendants whose alcohol abuse problems are serious enough to result in felony DUI convictions.
Third, defendants with DUI convictions that are counted under ? 924(e)(2)(B) have either (a) such serious alcohol abuse problems that they have at least three prior felony DUI convictions or (b) both one or two felony DUI convictions and one or two offenses that fall under ? 924(e)(2)(B)(i) (offenses that have “as an element the use, attempted use, or threatened use of physical force”) or that are specifically set out in ? 924(e)(2)(B)(ii) (burglary, arson, extortion, or an explosives offense). Defendants with three felony DUI convictions are likely to be super-DUI-recidivists like petitioner. Defendants with a combination of felony DUI and other qualifying convictions?for example, convictions for assault or burglary?are persons who, even by the Court’s lights, could be classified as “the kind of person who might deliberately point [a] gun and pull the trigger.”
Unlike the Court, I cannot say that persons with these characteristics are less likely to use a gun illegally than are persons convicted of other qualifying felonies.
Justice SCALIA’s concurrence takes a different approach, but his analysis is likewise flawed. Justice SCALIA would hold (a) that an offense does not fall within the residual clause unless it presents a risk that is at least as great as that presented by the least dangerous of the enumerated offenses; (b) that burglary is the least dangerous of the enumerated offenses; (c) that the relevant measure of risk is the risk that the typical burglary, DUI, etc. would result in injury; and (d) that the risk presented by an incident of DUI is less than the risk presented by a burglary.
Justice SCALIA, like the Court, does not follow the statutory language. The statute says that offenses falling within the residual clause must present “a serious potential risk of physical injury to another.” The statute does not say that these offenses must present at least as much risk as the enumerated offenses.
The statute also does not say, as Justice SCALIA would hold, that the relevant risk is the risk that each incident of DUI will result in injury. I see no basis for concluding that Congress was not also concerned with the risk faced by potential victims, particularly since the statute explicitly refers to “potential risk.” Drunk driving is regarded as a severe societal problem in large measure because of the very large number of victims it produces each year.
Finally, Justice SCALIA’s conclusion that burglary is the least risky of the enumerated offenses is based on a procrustean reading of ? 924(e)(2)(B)(ii). This provision refers, without qualification, to “extortion.” In his dissent in James v. United States, 550 U.S. ___, 127 S. Ct. 1586 (2007), Justice SCALIA concluded that many forms of extortion are “inherently unlikely to cause physical harm.” Id., at ___, 127 S.Ct., at 1594-95 (emphasis in original). Only by finding that the *1597 term “extortion” in ? 924(e)(2)(B)(ii) really means only certain forms of extortion was Justice SCALIA able to come to the conclusion that burglary is the least risky of the enumerated offenses.
For all these reasons, I would affirm the decision of the Tenth Circuit.
 United States v. Gonzaga Rodriquez, now pending before the Court, presents the question “[w]hether a state drug-trafficking offense, for which state law authorized a ten-year sentence because the defendant was a recidivist, qualifies as a predicate offense under the Armed Career Criminal Act, 18 U.S.C. ? 924(e).” Pet. for Cert., O.T.2007, No. 06-1646, p. I.
 See the National Highway Traffic Safety Administration (NHTSA) Traffic Safety Facts Ann. Rep., p. 56, Table 34 (2006) (15,945 alcohol-related fatal crashes; 41%), (2005) (15,238; 39%), (2004) (14,968; 39%), (2003) (15,251; 40%), (2002) (15,626; 41%), (2001) (15,585; 41%), (2000) (14,847; 40%), (1999) (14,109; 38%), (1998) (14,278; 39%), (1997) (14,363; 38.5%), (1996) (15,249; 40.8%) online at http://www-nrd.nhtsa.dot.gov/ CMSWeb/listpublications.aspx?Id=E&Show By=DocType (all Internet materials as visited Apr. 11, 2008, and available in Clerk of Court’s case file); see also Michigan Dept. of State Police v. Sitz, 496 U.S. 444, 451, 110 S. Ct. 2481, 110 L. Ed. 2d 412 (1990) (“No one can seriously dispute the magnitude of the drunken driving problem …. `Drunk drivers cause an annual death toll of over 25,000 and in the same time span cause nearly one million personal injuries …'”) (footnote omitted); South Dakota v. Neville, 459 U.S. 553, 558, 103 S. Ct. 916, 74 L. Ed. 2d 748 (1983) (“The carnage caused by drunk drivers is well documented …. This Court … has repeatedly lamented the tragedy”).
 See NHTSA, supra, at 111, Table 76 (2006) (278,000), (2005) (254,000), (2004) (248,000), (2003) (275,000), (2002) (258,000), (2001) (275,000), (2000) (310,000), (1999) (308,000), (1998) (305,000), (1997) (327,000), (1996) (321,000).
 According to statistics compiled by the Federal Bureau of Investigation, between 1996 and 2006 total annual murders never exceeded 15,000 after 1997. During that same 11-year period, the highest number of murders committed in the course of burglary was 123, the number of murders committed in the course of arson peaked at 105, and the number of murders involving explosives topped out at 14?all in 1996. See Dept. of Justice, Federal Bureau of Investigation, Uniform Crime Reports/Crime in the United States (Ann.Reps. 1996-2006), online at http://www. fbi.gov./ucr/ucr.htm#cius. While murders committed in the course of extortion were not separately reported, common sense and the fact that the total number of murders was similar to the number of fatal alcohol-related crashes at least after 1997 indicates that murders involving extortion would not rival deaths in alcohol-related auto accidents. Even if one were to expand beyond murders to all fatalities and even injuries, it is estimated that arson causes the relatively small number of 475 deaths and over 2,000 injuries annually. Dept. of Homeland Security, U.S. Fire Administration, Arson in the United States, Vol. 1 Topical Fire Research Series, No. 8 (Jan.2001, rev.Dec.2001), online at http://www.usfa.dhs.gov/downloads/pdf/tfrs/vli 8-508.pdf.
 See United States v. McCall, 439 F.3d 967, 972 (C.A.8 2006) (en banc) (citing Brewer et al., The Risk of Dying in Alcohol-Related Automobile Crashes Among Habitual Drunk Drivers, 331 New Eng. J. Med. 513 (1994)); Dept. of Justice, Office of Community Oriented Policing Services, Drunk Driving, Problem-Oriented Guides for Police, Problem-Specific Guides Series No. 36, p. 4 (Feb.2006) (“By most estimates, although repeat drunk drivers comprise a relatively small proportion of the total population of drivers, they are disproportionately responsible for alcohol-related crashes and other problems associated with drunk driving”).
 Brewer, supra, text accompanying nn. 23-24; L. Taylor & S. Oberman, Drunk Driving Defense ? 1.01 (2007).
 See National Conference of State Legislatures, Criminal Status of State Drunk Driving Laws, online at http://www.ncls.org/ programs/lis/dui/felonn.htm (current as of July 2003) (surveying 50 States, the District of Columbia, and U.S. Territories, most of whom treat the first DUI offense as a misdemeanor).