dissenting.
Wenner pled guilty to burglarizing “a dwelling other than a vehicle, residence of Mike Jewell.” The majority concludes that Wenner did not commit a “burglary of a dwelling” or any other “crime of violence” as defined under the Sentencing Guidelines. I dissent from the majority’s strained interpretation of federal law.
The district court should be affirmed for two reasons. First, while we use the categorical approach outlined in Taylor v. United States, 495 U.S. 575, 110 S.Ct. 2143, 109 L.Ed.2d 607 (1990), United States v. Becker, 919 F.2d 568, 570 (9th Cir.1990) (applying Taylor’s categorical approach to the Guidelines), we do not extend Taylor’s definition of generic burglary to the Guidelines’ more specific offense “burglary of a dwelling.” Second, the district court should be affirmed because, even if Washington defines residential burglary more broadly than the Guidelines’ “burglary of a dwelling,” residential burglary is nonetheless a crime of violence under the Guidelines’ catch-all provision.
I.
Wenner was convicted in state court of residential burglary, which Washington defines as “entering] or remaining] unlawfully in a dwelling other than a vehicle” with the intent to commit a crime. Wash. Rev. Code § 9A.52.025(1). A “dwelling” is “any building or structure, though movable or temporary, or a portion thereof, which is used or ordinarily used by a person for lodging.” Id. § 9A.04.110(7). In Washington, a “building” can include a fenced area, a railway car, or a cargo container. Id. § 9A.04.110(5).
Wenner’s conviction is a “crime of violence” under U.S. Sentencing Guidelines § 4B1.2(a)(2) if it represents “an[ ] offense under federal or state law, punishable by imprisonment for a term exceeding one year, that ... is burglary of a dwelling ... or otherwise involves conduct that presents a serious potential risk of physical injury to another.” To decide whether Wenner’s conviction constitutes a “burglary of a dwelling” under federal law, we must compare Washington’s crime of “burglary of a residence” with “burglary of a dwelling” under section 4B1.2(a)(2), looking exclusively to the statutory definition of the crime rather than to the specific conduct underlying Wenner’s conviction. Taylor, 495 U.S. at 602, 110 S.Ct. 2143. If the state statute criminalizes conduct that would not be a crime of violence under section 4B1.2(a)(2), Wenner’s conviction cannot support the higher base offense level unless “the jury was actually required to find all the elements” of a crime of violence. Taylor, 495 U.S. at 602, 110 S.Ct. 2143.
II.
Wenner’s argument proceeds as follows. First, a “burglary of a dwelling” under the Guidelines must be defined the same way Taylor defined “burglary” under the Armed Career Criminal Act (ACCA), i.e., it must involve entry into a building. 495 U.S. at 598, 110 S.Ct. 2143. Second, the federal definition of “building” is narrower than Washington’s definition of a “dwelling.” Consequently, some things deemed residences under Washington law (e.g., fenced areas, railway cars, and cargo containers, Wash. Rev. Code § 9A.52.025(1), 9A.04.110(5)) are not buildings or structures under federal law and so cannot support a conviction for generic “burglary” under Taylor. See United States v. Bonat, 106 F.3d 1472, 1477 (9th Cir.1997) (burglary of a railway car is not “burglary” under Taylor and the ACCA); United States v. Pluto, 144 F.3d 968, 975-76 (6th Cir.1998) (burglary of neighbor’s backyard is not “burglary” under Taylor and the ACCA). *978In sum, Wenner contends that the Washington’s residential burglary statute is broader than section 4B1.2(a)(2), because burglarizing a fenced area that doubles as a dwelling is a residential burglary under Washington law, but not a “burglary” under Taylor, and therefore not a burglary of a dwelling under the Guidelines.
Wenner assumes, and the majority agrees, that Taylor’s definition of “burglary” applies to the Guidelines’ offense “burglary of a dwelling,” and thus a burglary of a dwelling under the Guidelines must involve a “building or structure” under Taylor. I do not accept this conclusion. Taylor defined run-of-the-mill burglary as that word appears in the ACCA, 18 TJ.S.C. § 924(e) (2) (B) (ii); we do not apply Taylor’s general definition to more specific types of burglary.
Suppose, for instance, the Sentencing Commission were to add a provision to the Guidelines to increase a defendant’s base offense level for a vehicular burglary conviction. Surely Taylor’s definition of generic “burglary” would not apply, else the vehicular burglary would have to transpire inside a building. See Sareang Ye v. INS, 214 F.3d 1128, 1132-33 (9th Cir.2000) (recognizing that Taylor’s general definition of “burglary” does not include “vehicle burglary”). A person who stole a car from a fenced area (such as a parking lot) would not have committed a vehicular burglary. This absurd result is just what the majority’s logic commands.
Reason compels me to conclude that Taylor’s definition of “burglary” does not apply every time that word appears in the Sentencing Guidelines. Taylor does not purport to define “burglary of a dwelling” as that term is employed in section 4B1.2(a)(2). Thus, our general obligation to apply Taylor’s categorical approach when interpreting the Sentencing Guidelines does not require us to extend its generic definition blindly to the Guidelines’ more specific “burglary of a dwelling.”
The issue before us is not whether residential burglary is “burglary” under the ACCA. It is not whether residential burglary satisfies Taylor’s definition of “burglary” or whether fenced areas are “buildings” under federal law. Instead, the issue is whether Washington’s residential burglary statute is broader than the federal definition of “burglary of a dwelling.” More precisely, we must decide whether the burglary of fenced areas, railway cars, or cargo containers used for lodging constitutes “burglary of a dwelling” under the Guidelines. To this I turn.
The Third and Eighth Circuits define a “dwelling” for these purposes as a “building or portion thereof, a tent, a mobile home, a vehicle or other enclosed space which is used or intended for use as a human habitation, home or residence.” United States v. McClenton, 53 F.3d 584, 587 (3d Cir.1995), quoting Black’s Law Dictionary 505 (6th ed.1990); United States v. Graham, 982 F.2d 315, 316 (8th Cir.1992) (quoting same). If we interpret “burglary of a dwelling” to cover any “enclosed space which is used or intended for use as a human habitation, home, or residence,” Washington’s residential burglary statute is no broader than the uniform federal definition. A fenced area is an “enclosed space,” as are railway cars and cargo containers. See, e.g., Black's Law Dictionary 547, 881 (7th ed.1999) (defining “enclosed land” as “land that is actually enclosed and surrounded by fences”); 29 C.F.R. § 1926.21(b)(6)(h) (defining “enclosed space” for safety and health regulations as including “open top spaces more than four feet in depth”). Rather than embrace the majority’s illogical conclusion that generic “burglary” and “burglary of a dwelling” must both involve buildings, I *979would hold that Washington’s residential burglary is no broader than 4B1.2(a)(2) and thus satisfies Taylor’s categorical approach.
The majority asserts that its definition of “burglary of a dwelling” does not create a circuit split because McClenton and Graham merely narrow Taylor’s definition for purposes of section 4B1.2(a)(2) by requiring entry into a building that is also a dwelling. Although Graham is somewhat ambiguous on this point, McClenton’s plain language defies the majority’s construction: “The Sentencing Commission has adopted a categorical approach to the determination of whether an underlying offense is a ‘crime of violence’ within section 4B1.2, deciding that any invasion of a place where people may reside presents an unacceptable risk of harm and must be classified as a crime of violence.” McClenton, 53 F.3d at 588. By defining “burglary of a dwelling” differently from Taylor’s definition for generic burglary, McClenton directly contradicts the majority’s decision.
In the alternative, the majority contends that McClenton’s interpretation of section 4B1.2(a)(2) cannot be correct because “the Supreme Court has rejected the argument that ‘burglary’ under the ACCA should be construed to apply to a narrow subclass of burglaries.” Contrary to the majority’s assertion, Taylor did not foreclose the possibility that the Sentencing Commission might depart from its general definition of “burglary” to identify more specific categorizations. In reality, the Third Circuit’s generic definition of “burglary of a dwelling” is both broader and narrower than the Sentencing Guidelines’ definition for run-of-the-mill “burglary”: it is broader than Taylor’s definition, because it applies to residential spaces that are not buildings, but it is also considerably narrower, because it does not apply to nonresidential buildings. Nowhere in Taylor does the Court suggest that the Sentencing Commission lacks the authority to delimit a more . specific burglary category along these lines that departs from Taylor’s generic definition.
The majority also argues that a “residence” under federal law must be a “building” under United States v. Williams, 47 F.3d 993 (9th Cir.1995). Although Williams does include a reference to Thompson’s “building or structure” language, it does not address the issue raised in this case. In Williams, we cited Thompson as support for our holding that Williams’s sentence could be enhanced notwithstanding the fact that the indictment did not employ the specific terms “unlawful or unprivileged entry” as required by United States v. Parker, 5 F.3d 1322, 1325 & n. 2 (9th Cir.1993). Because Williams’s burglary took place in an “occupied structure,” we did not have occasion to consider whether illegal entry into an alternative residential space would constitute “burglary of a dwelling.” Thus, we did not consider whether Taylor’s general definition of “burglary” restricts the ordinary meaning of “residence” as used in section 4B1.2(a)(2).
Given that the majority’s opinion draws this circuit into conflict with the Third Circuit, I find it highly ironic that the majority faults my reading of section 4B1.2(a)(2) for undermining the Guidelines’ goal “to promote uniformity in sentencing.” The majority’s holding frustrates inter-circuit sentencing uniformity by unnecessarily extending a generic definition of “burglary” to a context where the Sentencing Guidelines clearly anticipate a definition that would apply to all residential spaces. For these reasons, I would interpret “burglary of a dwelling” under section 4B1.2(a)(2) more broadly to encompass the residential spaces covered in Washington’s residential burglary statute.
*980III.
Regardless of my conflict with the majority above, there is an alternative theory upon which we can and should affirm the district court. I now turn to this alternative.
The majority argues that we may not consider whether Wenner’s acts “otherwise involve[ ] conduct that presents a serious potential risk of physical injury to another” because the parties did not brief this issue and the district court did not apply section 4B1.2(a)(2)’s catch-all provision. The sentencing court stated:
I appreciate the argument that defendant makes narrowing- — trying to narrow — these things to get around what is apparently the intent of the statute, or the guideline, but I frankly just don’t buy it. I think if we look at the ordinary terms used and the statutes and what Mr. Wenner in fact did based on the records of those cases, I think that both the residential burglary and the attempted residential burglary are crimes of violence — not actual violence, but violence as that term is misused in the guidelines. That may be unfortunate, but I believe that the government simply has the better side of that argument, and so I think the guideline workup is correct.
Although the sentencing court’s rationale is not crystal clear, its decision to “look at the ordinary terms used[and] what Mr. Wenner in fact did based on the records of those cases” suggests that it likely considered, at least in part, applying “burglary of a dwelling” and relied on section 4B1.2(a)(2)’s catch-all provision.
But this is a non-dispositive issue. Whether the sentencing court in fact relied on the catch-all provision is not determinative for purposes of our analysis. Because the Sentencing Guidelines’ applicability to a particular offense is a question of law we review de novo, United States v. Alcarez Camacho, 340 F.3d 794, 796 (9th Cir.2003), we are not bound to decide the case based solely on the sentencing court’s rationale or the district court’s reason for dismissing the habeas petition. We can affirm if the record shows the sentencing court got it right but for another reason. See id. (‘Whether the sentencing guidelines apply to an offense is a question of law reviewed de novo, without deference to the sentencing court’s interpretation.” (internal quotations and citation omitted)); Pollard v. White, 119 F.3d 1430, 1433 (9th Cir.1997) (“We review a district court’s dismissal of a habeas petition de novo and may affirm on any ground supported by the record, even if it differs from the rationale of the district court.”). Therefore, even assuming that Wenner’s residential burglary was not a burglary of a dwelling under section 4B1.2(a)(2), I believe we should affirm the district court under the section’s catch-all provision, because the conviction involves “conduct that presents a serious potential risk of physical injury to another.”
Under Washington law, Wenner’s conviction necessarily involved a delimited space “used or ordinarily used by a person for lodging.” Wash. Rev. Code § 9A.04.110(7). Our prior decisions dictate that this offense presents a serious potential risk of physical injury to another. Thus, Washington’s residential burglary statute is no broader than the Guidelines’ catch-all provision.
For a state crime to be a crime of violence under section 4B1.2(a)(2)’s catchall provision, we need not — and cannot— look to the particular facts that occasioned the conviction. Rather, we must apply the categorical approach, under which we look only at the statutory definition. United States v. M.C.E., 232 F.3d 1252, 1255 (9th Cir.2000). We need not consider whether the residence Wenner burglarized was ac*981tually occupied during the burglary. United States v. Weinert, 1 F.3d 889, 891 (9th Cir.1993). All that matters is that the state statute only criminalizes conduct that presents a serious potential risk of physical injury to another.
The federal statutory provision interpreted in M.C.E. defines a crime of violence to include any crime that “by its very nature, involves a substantial risk that physical force against the person of another may be used in committing the offense.” 18 U.S.C. § 5032. M.C.E. analyzed the same Washington statute at issue here and concluded that it was a crime of violence under 18 U.S.C. § 5032 “because the perpetrator’s unlawful entry into a dwelling with intent to commit a crime therein creates a substantial risk that he may encounter the lawful occupant, or perhaps an investigating police officer, thus resulting in a violent confrontation.” 232 F.3d at 1255. Notwithstanding differences in state definitions, “[cjourts that have faced this question have come to the conclusion (unanimously as far as we can tell) that residential burglary is indeed a crime of violence.” Id. Such conduct is inherently dangerous, not because it may take place in a “structure or building,” but rather because violence is highly likely to occur if the perpetrator encounters a lawful occupant or police officer. Id.; see United States v. Pinto, 875 F.2d 143, 144 (7th Cir.1989) (“[R]esidential burglary has been considered a violent offense for hundreds of years.... No one has doubted for decades that residential burglary is a ‘violent offense, because of the potential for mayhem if the burglar encounters a resident.’ ”); United States v. Davis, 881 F.2d 973, 976 (11th Cir.1989) (“In accord with ... the settled law of the federal circuits, we conclude that the burglary of a dwelling by its nature creates a substantial risk of physical force.”). M.C.E. ’s reasoning is even more forceful in the instant case since 18 U.S.C. § 5032 required a “substantial risk,” while section 4B 1.2(a)(2) requires only a “serious potential risk.” Accord Becker, 919 F.2d at 571-72 (daytime burglary of a residence is a crime of violence under the 1988 Guidelines, which defined a crime of violence as including a felony that inherently involves “a substantial risk” of physical force). M.C.E. therefore compels the conclusion that Wenner’s conviction under Washington’s residential burglary statute constitutes a crime of violence under the Guidelines’ catch-all provision.
It is no response to assert that this reading of the catch-all provision would render the more specific provision “burglary of a dwelling” mere surplusage. Catchall provisions are designed to catch what specific provisions leave behind. At most, section 4B1.2(a)(2)’s catch-all provision may be subject to the ejusdem generis canon, which dictates that “[wjhere general words follow the enumeration of specific classes of things, the general words must be construed as restricted to things of the same type as those specifically enumerated.” Aqua-Marine Constructors, Inc. v. Banks, 110 F.3d 663, 677 (9th Cir.1997); citing 2A Norman J. Singer, Sutherland Statutory Construction § 47.17, at 188-90 (5th rev. ed.1992); see also Wash. Dep’t of Social and Health Serv. v. Estate of Keffeler, 537 U.S. 371, 384-85, 123 S.Ct. 1017, 154 L.Ed.2d 972 (2003) (applying this principle). Ejusdem generis actually supports my reading of section 4B1.2(a)(2), however. As the section’s commentary explains, an unlisted offense is a crime of violence (i.e., of “the same type as those specifically enumerated”) if the conduct “by its nature, presented a serious potential risk of physical injury.” U.S.S.G. § 4B1.2, cmt. 1. The majority does not seriously dispute that illegally entering or remaining in a fenced area, railway car, or cargo container used as a residence with intent to commit a *982crime therein poses a serious risk of physical injury, whether or not this conduct constitutes “burglary of a dwelling” under federal law.
Reading section 4B1.2(a)(2)’s catch-all provision to include this conduct does not render the statute’s specific reference to “burglary of a dwelling” superfluous. On the contrary, “burglary of a dwelling” provides the paradigmatic example against which we must evaluate whether Wenner’s conviction is a crime of violence. Burglary of a fenced area, railway car, or cargo container used as a residence constitutes a crime of violence because it raises precisely the same safety concerns raised by other types of residential burglary. This common sense analogical approach to section 4B 1.2(a)(2) is consistent with Taylor and tracks our sister circuits’ recent holdings. See Taylor, 495 U.S. at 599 n. 9, 110 S.Ct. 2143 (“The Government remains free to argue that any offense — including offenses similar to generic burglary — should count toward enhancement as one that ‘otherwise involves conduct that presents a serious potential risk of physical injury to another’.... ”); United States v. Sun Bear, 307 F.3d 747, 753 (8th Cir.2002) (deciding that vehicle theft meets this test).
In sum, ejusdem generis confirms the conclusion that Wenner’s state law conviction involves conduct that poses a “serious potential risk of physical injury to another.” That is, even if Wenner did not technically burglarize a “dwelling” under the majority’s ill-conceived definition, he still committed a crime of violence under the catch-all provision. To conclude otherwise is to disregard the catch-all provision’s clear purpose and improperly restrict the provision’s scope.
IV.
The majority incorrectly defines “burglary of a dwelling” under the Guidelines and impermissibly neglects section 4B1.2(a)(2)’s catch-all provision. Under both tests, the district court should be affirmed in concluding that Wenner’s sta-telaw conviction for residential burglary constitutes a crime of violence. I therefore dissent.