concurring in the judgment.
I disagree with the Court’s interpretation of 18 U. S. C. § 1512(a)(1)(C). In my view, the Government must prove *679that the defendant intended to prevent a communication which, had it been made, would beyond a reasonable doubt have been made to a federal law enforcement officer. The Court’s vague “reasonable likelihood” standard has no basis in the statutory text and will serve only to confuse judges and juries. Accordingly, although I agree the case should be remanded for the Eleventh Circuit to consider whether the objection to sufficiency of the evidence was preserved or whether the District Court committed plain error, I would hold that there was insufficient evidence to support Fowler’s conviction.
I
Section 1512(a)(1)(C) of Title 18 makes it a federal crime “to kill another person, with intent to . . . prevent the communication by any person to a law enforcement officer ... of the United States of information relating to the commission or possible commission of a Federal offense.” Viewed in isolation, this provision contains an ambiguity: Does the mens rea of the statute include a specific intent to prevent communication to a law enforcement officer of the United States; or is it satisfied by the mere intent to prevent communication to a law enforcement officer who happens to be a law enforcement officer of the United States?
Happily, a different statutory provision resolves this ambiguity. It states that “no state of mind need be proved with respect to the circumstance . . . that the law enforcement officer is an officer or employee of the Federal Government.” § 1512(g)(2). This makes clear that the first possibility is wrong, and the second right. But removing the “federal officer” requirement as an element of the statute’s mens rea does not remove it as an element of the actus reus — that is, as an element of the facts that must be proved for conviction. It must be proved, and proved beyond a reasonable doubt, that the communication intended to be prevented was communication to a federal officer. Thus, if a suspect in an investigation murders an informant to prevent him from talk*680ing to authorities, but is unaware that the informant was working for the FBI, the suspect would be guilty: He would have committed a murder with the intention of preventing the informant’s communication to authorities about his criminal activities, and the communication he sought to prevent would necessarily have been to federal law enforcement. Likewise, a suspect would be guilty if he committed a murder to prevent a witness from informing law enforcement that he lied on his federal income tax return: He sought to prevent a communication that would have been made to federal officials, because they alone prosecute federal tax violations. But a suspect who commits a murder with the general intent of preventing law enforcement from learning about activities that violate both state and federal law would not be . guilty, because the Government would be unable to prove that the communication he sought to prevent necessarily would have been to a federal official.
Applying that standard, this is an easy case. There was evidence that Fowler murdered Officer Horner in order to prevent him from communicating information about Fowler’s criminal activities. But the only evidence proffered by the Government to establish that the communication would have been to a federal law enforcement agent was the fact that a different state police officer, four years later, contacted federal law enforcement about a robbery by Fowler’s confederate — and that only because the state-law statute of limitations for the robbery had expired. That is not nearly enough to demonstrate Fowler’s guilt beyond a reasonable doubt.
II
The Court gives the statute a broader reading than the one I ascribe. The Government can obtain a conviction, it says, so long as it can prove a “reasonable likelihood” that the communication would have been made to a federal law enforcement officer. I know of no precedent for using a “likelihood” standard rather than the “beyond a reasonable *681doubt” standard for a finding of fact essential to a criminal conviction; and the justifications the Court presents for that course in the present case are not convincing.
The Court maintains that the Government need not show beyond a reasonable doubt that the communication would have been to a federal officer because “[t]he relevant question concerns the defendant’s intent.”' Ante, at 674. But that reasoning is directly contrary to § 1512(g)(2), which expressly states that the defendant’s intent is not the relevant question with respect to the federal character of law enforcement officer meant to be deprived of the information. The Court’s observation that “a defendant can kill a victim with an. intent to prevent the victim from communicating with federal law enforcement officers even if there is some considerable doubt that any such communication would otherwise have taken place,” ibid., is completely irrelevant to the question presented.
The Court also proclaims that a narrower view “would conflict with the statute’s basic purpose,” which is to prevent witness tampering “at a time when the precise communication and nature of the officer who may receive it are not yet known.” Ante, at 673. It cites no basis for attributing that purpose, and there is none — other than the fact that it supports the Court’s outcome. Another purpose is just as likely — and indeed more likely, since it can be achieved without abandonment of the ancient rule that in criminal prosecutions facts must be found beyond a reasonable doubt. Murder, after all, is a crime, and often a capital crime, under all state laws. There is no reason to ascribe to Congress the “purpose” of transferring murder prosecutions that would ordinarily be brought in state court to federal court based on only a tangential federal interest. Congress was concerned with preserving the integrity and effectiveness of federal prosecutions, and where they are not clearly involved (as the ordinary beyond-a-reasonable-doubt standard would require) a federal murder prosecution has no proper place. Limited *682as I have suggested, the federal law would still have ample scope, reaching what were surely the principal cases Congress had in mind — the killing of prospective witnesses in federal trials or in ongoing federal investigations. Here, as would be the case in many situations involving a merely hypothetical link to a federal investigation, Fowler murdered a state police officer. The natural place to have prosecuted him would have been state court.
The Court’s analysis is even less persuasive in light of the rule of lenity, under which we must construe ambiguous criminal statutes in favor of the defendant. Here, the Court adopts a kind of rule of harshness, discarding the most straightforward construction of the text in favor of a textually implausible one, based on vague intuitions about the statute’s purpose. The Court’s opinion never cites the rule of lenity, probably because it cannot honestly say that the statute is so clear that “there is no ambiguity for the rule of lenity to resolve.” Burgess v. United States, 553 U. S. 124, 136 (2008).
To make matters worse, the Court’s standard is hopelessly indeterminate. The Government must show that a communication to a federal officer is “reasonably likely,” which is less likely than “more likely than not,” but more likely than “reasonably possible.” Ante, at 675-678. I doubt that any jury can grasp the distinction between “you must find that a communication to a federal officer was reasonably likely” and “you must find that a communication to a federal officer was reasonably possible.” Understandably, the Court refuses to give any examples of what “reasonably likely” means, except for an absurd example involving communications with Lithuanian police officers, ante, at 678 — which obviously would not be “reasonably possible” either. Indeed, the Court refuses to apply its standard to the facts of this case, leaving that precarious task to the lower court.
Ill
The dissent adopts a view of the statute that is even broader than the Government’s. It effectively contends that *683the Government need not prove anything with respect to the fact that the communication sought to be prevented was “to a law enforcement officer ... of the United States.” As long as the Government can prove that the defendant sought to prevent the communication of information about a federal crime (including a federal crime that is also a state crime) it will necessarily have proved that the “set of law enforcement officers (whose identities were unknown to him)” he had in mind “included law enforcement officers who were employed by the United States.” Post, at 687 (opinion of Alito, J.). Conviction requires neither any specific intent regarding the federal status of the officer, nor even any likelihood that a commimication to a federal officer would have occurred.
The principal defect in this interpretation is that it makes the words “of the United States” superfluous. Section 1512(a)(1)(C) specifically requires that the information the defendant seeks to prevent from being communicated be “information relating to the commission or possible commission of a Federal offense.” If the phrase “to a law enforcement officer ... of the United States” requires nothing more than this it is utterly without effect. The implication of this view is that Congress enacted § 1512(a)(l)(C)’s reference to “a law enforcement officer... of the United States,” only to immediately nullify it by § 1512(g)(2)’s “no state of mind” provision. Not likely — and not sound statutory interpretation.
The dissent claims that my analysis “confuses what the prosecution must prove with what a rational jury may choose to infer in a particular case.” Post, at 690. I find this contention difficult to understand. In the dissent’s view, a properly instructed jury should be required to find neither that the defendant’s mens rea had any connection to a federal officer, nor that the defendant’s actus reus had any connection to a federal officer. It therefore follows that under the dissent’s view, a properly instructed jury should be required to find nothing about a connection to a federal officer beyond the fact that the information related to a federal offense, which means that, unless the jury is acting irrationally or *684is engaging in jury nullification, the “of the United States” provision is indeed superfluous. The dissent is correct that the proof of one element of a crime (such as an overt act) can sometimes be used to prove that a different element (such as a conspiratorial agreement) is satisfied, ibid., n. 2; but in such cases, the jury is instructed that it is required to make a separate finding to convict (e. g., that a conspiratorial agreement actually occurred). Here, the dissent identifies no separate finding the jury must make beyond the fact of a federal offense. The dissent also observes that when a defendant murders a federal officer to prevent him from communicating information about a nonfederal crime, he does not violate the statute. Post, at 691. This observation convincingly establishes that the statutory words “Federal offense” are not superfluous under the dissent’s view, an observation irrelevant to my point that the dissent makes the statutory words “of the United States” superfluous.
The dissent contends that my interpretation “has no grounding in the language of the statute.” Post, at 688. It asserts that “the text of the statute makes it perfectly clear that the federal officer requirement is exclusively an element of the defendant’s mens rea.” Post, at 689 (internal quotation marks omitted). Perhaps the only thing “perfectly clear” about this statute is that it states the precise opposite of that proposition: “[N]o state of mind need be proved with respect to the circumstance . . . that the law enforcement officer is an officer or employee of the Federal Government!” § 1512(g)(2).
The dissent’s interpretation would federalize crimes that have no connection to any federal investigation. A person caught by a state police officer with marijuana who murders the state police officer to cover it up could be prosecuted in federal court. That would approach the outer limits of Congress’s enumerated powers. We have adopted a federalism principle that applies when a statute would render “traditionally local criminal conduct... a matter for federal en*685forcement”: “[Ujnless Congress conveys its purpose clearly, it will not be deemed to have significantly changed the federal-state balance in the prosecution of crimes.” Jones v. United States, 529 U. S. 848, 858 (2000) (internal quotation marks omitted). Thus, the dissent adds to the Court’s “rule of harshness” a rule of antifederalism, under which a court must actually ignore a federal connection that Congress prescribed so as to avoid intrusion into traditionally local law enforcement.
Because the Government did not establish that Fowler intended to prevent a communication that, if made, would have been made to a federal law enforcement officer, there was insufficient evidence to convict him of violating § 1512(a)(1)(C). Since there remains, however, the question whether Fowler preserved this issue at trial or whether the inadequacy of the evidence constituted plain error, I concur in the Court’s order vacating the judgment and remanding for resolution of that question.