with whom Justice Souter joins, concurring.
I agree with the Court’s holding that a statute, “found to be civil [in nature], cannot be deemed punitive” or criminal “as applied” for purposes of the Ex Post Facto and Double Jeopardy Clauses. Ante this page. The Court accurately observes that this holding gives us “no occasion to consider ... the extent to which a court may look to actual conditions of confinement and implementation of the statute to determine in the first instance whether a confinement scheme is civil in nature.” Ante, at 266. I write separately to dissociate myself from any implication that this reserved point may be an open question. I do not regard it as such since, three *268years ago, we rejected a similar double jeopardy challenge (based upon the statute’s implementation “as applied” to the petitioners), where the statute had not yet been determined to be civil in nature, and where we were making that determination “in the first instance.” See Hudson v. United States, 522 U. S. 93 (1997). To be consistent with the most narrow holding of that case (which, unlike this one, did not involve imposition of confinement), any consideration of subsequent implementation in the course of making a “first instance” determination cannot extend to all subsequent implementation, but must be limited to implementation of confinement, and of other impositions that are “not a fixed event,” ante, at 263. That, however, would be a peculiar limitation, since even “fixed events” such as the imposition of a fine can, in their implementation, acquire penal aspects— exemplified in Hudson by the allegedly punitive size of the fines, and by the availability of reduction for “good-faith” violations, see 522 U. S., at 97-98,104. Moreover, the language and the reasoning of Hudson leave no room for such a peculiar limitation.
In that case, the petitioners contended that the punitive nature of the statute that had been applied to them could be assessed by considering the aforementioned features of the fines. We flatly rejected that contention, which found support in our prior decision in United States v. Halper, 490 U. S. 435 (1989). Halper, we said, had erroneously made a “significant departure” from our prior jurisprudence, in deciding “to ‘asses[s] the character of the actual sanctions imposed,’ 490 U. S., at 447, rather than, as Kennedy [v. Mendoza-Martinez, 372 U. S. 144 (1963),] demanded, evaluating the ‘statute on its face’ to determine whether it provided for what amounted to a criminal sanction, [id.], at 169.” 522 U. S., at 101. The Kennedy factors, we said, “ ‘must be considered in relation to the statute on its face,’ ” 522 U. S., at 100, quoting from Kennedy v. Mendoza-Martinez, 372 U. S. 144, 169 (1963). We held that “[t]he fact that petitioners’ *269'good faith’ was considered in determining the amount of the penalty to be imposed in this case [a circumstance that would normally indicate the assessment is punitive] is irrelevant, as we look only to 'the statute on its face’ to determine whether a penalty is criminal in nature.” Hudson, supra, at 104, quoting Kennedy, supra, at 169. We repeated, to be sure, the principle that the statutory scheme would be criminal if it was sufficiently punitive “ 'either in purpose or effect,’” Hudson, supra, at 99 (emphasis added), quoting United States v. Ward, 448 U. S. 242, 248-249 (1980), but it was clear from the opinion that this referred to effects apparent upon the face of the statute.
The short of the matter is that, for Double Jeopardy and Ex Post Facto Clause purposes, the question of criminal penalty vel non depends upon the intent of the legislature;* and harsh executive implementation cannot “ transform] what was clearly intended as a civil remedy into a criminal penalty,” Rex Trailer Co. v. United States, 350 U. S. 148, 154 (1956), any more than compassionate executive implementation can transform a criminal penalty into a civil remedy. This is not to say that there is no relief from a system that administers a facially civil statute in a fashion that would render it criminal. The remedy, however, is not to invalidate the legislature’s handiwork under the Double Jeopardy Clause, but to eliminate whatever excess in administration contradicts the statute’s civil character. When, as here, a state statute is at issue, the remedy for implementation that does not comport with the civil nature of the statute is resort to the traditional state proceedings that challenge unlawful executive action; if those proceedings fail, and the state *270courts authoritatively interpret the state statute as permitting impositions that are indeed punitive, then and only then can federal courts pronounce a statute that on its face is civil to be criminal. Such an approach protects federal courts from becoming enmeshed in the sort of intrusive inquiry into local conditions at state institutions that are best left to the State’s own judiciary, at least in the first instance. And it avoids federal invalidation of state statutes on the basis of executive implementation that the state courts themselves, given the opportunity, would find to be ultra vires. Only this approach, it seems to me, is in accord with our sound and traditional reluctance to be the initial interpreter of state law. See Railroad Comm’n of Tex. v. Pullman Co., 312 U. S. 496, 500-501 (1941).
With this clarification, I join the opinion of the Court.
Hudson v. United States, 522 U. S. 98 (1997), addressed only the Double Jeopardy Clause. Since, however, the very wording of the Ex Post Facto Clause — “No State shall... pass any ... ex post facto Law,” U. S. Const., Art. I, § 10, el. 1 (emphases added) — leaves no doubt that it is a prohibition upon legislative action, the irrelevance of subsequent executive implementation to that constitutional question is, if anything, even clearer.