concurring in the judgment.
The Court concludes that the National Traffic and Motor Vehicle Safety Act of 1966 (Safety Act) and Federal Motor Vehicle Safety Standard 208 (FMVSS 208) do not pre-empt the Williamsons’ state tort lawsuit. I agree. But I reach this result by a more direct route: the Safety Act’s saving clause, which speaks directly to this question and answers it. See 49 U. S. C. § 30103(e).
I
The plain text of the Safety Act resolves this case. Congress has instructed that “[e]ompliance with a motor vehicle safety standard prescribed under this chapter does not exempt a person from liability at common law.” Ibid. This saving clause “explicitly preserv[es] state common-law actions.” Wyeth v. Levine, 555 U. S. 555, 599 (2009) (Thomas, J., concurring in judgment). Here, Mazda complied with FMVSS 208 when it chose to install a simple lap belt. According to Mazda, the Williamsons’ lawsuit alleging that it should have installed a lap-and-shoulder seatbelt instead is pre-empted. That argument is foreclosed by the saving clause; the Williamsons’ state tort action is not pre-empted.
The majority does not rely on the Safety Act’s saving clause because this Court effectively read it out of the statute in Geier v. American Honda Motor Co., 529 U. S. 861 (2000). In Geier, the Court interpreted the saving clause as simply canceling out the statute’s express pre-emption clause with respect to common-law tort actions. This left the Court free to consider the effect of conflict pre-emption principles on such tort actions. See id., at 869.
But it makes no sense to read the express pre-emption clause in conjunction with the saving clause. See id., at 898 (Stevens, J., dissenting). The express pre-emption clause *340bars States from having any safety “standard applicable to the same aspect of performance” as a federal standard unless it is “identical” to the federal one. § 30103(b). That clause pre-empts States from establishing “objective rule[s] prescribed by a legislature or an administrative agency” in competition with the federal standards; it says nothing about the tort lawsuits that are the focus of the saving clause. Id., at 896.* Read independently of the express pre-emption clause, the saving clause simply means what it says: FMVSS 208 does not pre-empt state common-law actions.
II
As in Geier, rather than following the plain text of the statute, the majority’s analysis turns on whether the tort lawsuit here “ ‘stands as an obstacle to the accomplishment and execution of the full purposes and objectives’” of FMVSS 208. Ante, at 330 (quoting Hines v. Davidowitz, 312 U. S. 52, 67 (1941)). I have rejected purposes-and-objectives pre-emption as inconsistent with the Constitution because it turns entirely on extratextual “judicial suppositions.” Wyeth, supra, at 603 (opinion concurring in judgment); see also Haywood v. Drown, 556 U. S. 729, 756-759 (2009) (dissenting opinion).
Pre-emption occurs “by direct operation of the Supremacy Clause,” Brown v. Hotel Employees, 468 U. S. 491, 501 (1984), which “requires that pre-emptive effect be given only to those federal standards and policies that are set forth in, or necessarily follow from, the statutory text that was produced through the constitutionally required bicameral and presentment procedures.” Wyeth, 555 U. S., at 586 (opinion of Thomas, J.). In short, pre-emption must turn on *341the text of a federal statute or the regulations it authorizes. See id., at 587; see also Geier, supra, at 911 (Stevens, J., dissenting).
Purposes-and-objectives pre-emption — which by design roams beyond statutory or regulatory text — is thus wholly illegitimate. It instructs courts to pre-empt state laws based on judges’ “conceptions of a policy which Congress has not expressed and which is not plainly to be inferred from the legislation which it has enacted.” Hines, supra, at 75 (Stone, J., dissenting); Geier, supra, at 907 (opinion of Stevens, J.) (expressing concern about judges “running amok with our potentially boundless (and perhaps inadequately considered) [purposes-and-objectives pre-emption doctrine]”); see also Wyeth, supra, at 594-602 (opinion of Thomas, J.) (recounting the history of the doctrine).
The majority’s purposes-and-objectives pre-emption analysis displays the inherent constitutional problem with the doctrine. The Court begins with FMVSS 208, which allowed manufacturers to install either simple lap or lap-and-shoulder seatbelts in the rear aisle seat of 1993 minivans. The majority then turns to what it considers the primary issue: whether “that choice [was] a significant regulatory objective.” Ante, at 332 (emphasis added). Put more plainly, the question is whether the regulators really wanted manufacturers to have a choice or did not really want them to have a choice but gave them one anyway.
To answer that question, the majority engages in a “freewheeling, extratextual, and broad evaluatio[n] of the 'purposes and objectives’ ” of FMVSS 208. Wyeth, supra, at 604 (opinion of Thomas, J.). The Court wades into a sea of agency musings and Government litigating positions and fishes for what the agency may have been thinking 20 years ago when it drafted the relevant provision. After scrutinizing the 1989 Federal Register, a letter written in 1994, and the Solicitor General’s present-day assurances, the Court finds that the Department of Transportation liked the idea *342of lap-and-shoulder seatbelts in all seats, but did not require them, primarily for cost-efficiency reasons and also because of some concern for ingress-egress around the belt mounts. Ante, at 332-336. From all of this, the majority determines that although the regulators specifically and intentionally gave manufacturers a choice between types of seatbelts, that choice was not a “significant regulatory objective” and so does not pre-empt state tort lawsuits.
That the Court in Geier reached an opposite conclusion reveals the utterly unconstrained nature of purposes-and-objectives pre-emption. There is certainly “considerable similarity between this case and Geier. ” Ante, at 327. Just as in this case, Geier involved a choice offered to car manufacturers in FMVSS 208: whether to install airbags. Ante, at 332. And just as in this case, the Court in Geier relied on “history, the agency’s contemporaneous explanation, and the Government’s current understanding” to determine the significance of that choice. Ante, at 332. Yet the Geier Court concluded that “giving auto manufacturers a choice among different kinds of passive restraint devices was a significant objective of the federal regulation,” ante, at 330, and thus found the Geiers’ lawsuit pre-empted.
The dispositive difference between this case and Geier— indeed, the only difference — is the majority’s “psychoanalysis” of the regulators. United States v. Public Util. Comm’n of Cal., 345 U. S. 295, 319 (1953) (Jackson, J., concurring) (describing reliance on legislative history). The majority cites no difference on the face of FMVSS 208 between the airbag choice addressed in Geier and the seatbelt choice at issue in this case.
According to the majority, to determine whether FMVSS 208 pre-empts a tort suit, courts apparently must embark on the same expedition undertaken here: sifting through the Federal Register, examining agency ruminations, and asking the Government what it currently thinks. Pre-emption is then proper if the court decides that the regulators thought *343the choice especially important, but not if the choice was only somewhat important. This quest roves far from the Safety Act and analyzes pre-emption based on a formless inquiry into how strongly an agency felt about the regulation it enacted 20 years ago.
“[F]reeranging speculation about what the purposes of the [regulation] must have been” is not constitutionally proper in any case. Wyeth, 555 U. S., at 595 (opinion of Thomas, J.). The Supremacy Clause commands that the “[l]aws of the United States,” not the unenacted hopes and dreams of the Department of Transportation, “shall be the supreme Law of the Land.” U. S. Const., Art. VI, cl. 2. The impropriety is even more obvious here because the plain text of the Safety Act resolves this case.
For these reasons, I concur in the judgment.
See also Sprietsma v. Mercury Marine, 537 U. S. 51, 63-64 (2002) (addressing a similar express pre-emption clause and saving clause in the Federal Boat Safety Act, and holding that it is “perfectly rational” for Congress to bar state “administrative and legislative regulations” while allowing “private damages remedies” to compensate accident victims).