On direct appeal of a criminal conviction, the judgment can be affirmed despite a federal constitutional error only if the appellate court finds that the error was harmless beyond a reasonable doubt, the high court decided in Chapman v. California, 386 U. S. 18 (1967). Three years before AEDPA was enacted, the court decided that the limited purpose of habeas corpus required a higher hurdle to relief. A error would be considered harmless on habeas corpus unless it “had substantial and injurious effect or influence in determining the jury’s verdict.” See Brecht v. Abrahamson, 507 U.S. 619, 631 (1993).

The question before the court in Fry was which of these two standards applies when the state court did not do a harmless error analysis, usually because it found no error. Answer: Brecht.

In 1996, Congress added an additional prerequisite to overturning state criminal judgments on federal habeas corpus. If the state court resolved the issue on the merits, federal habeas relief cannot be granted unless that decision “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States. . . .” 28 U. S. C. § 2254(d)(1). If the state court did address the harmlessness issue, that decision stands unless the criteria of § 2254(d)(1) are met, and the Supreme Court confirmed this in Mitchell v. Esparza, 540 U. S. 12 (2003). However, the statute does not say, and Esparza does not hold, that the Brecht standard no longer applies. The Supreme Court rejected that argument in Fry, at p. 6.

So far, so good. This is holding, directly responsive to petitioner's argument. The holding is supported with this observation on the purpose of the statute:

Given our frequent recognition that AEDPA limited rather than expanded the availability of habeas relief, see, e.g., Williams v. Taylor, 529 U. S. 362, 412 (2000), it is implausible that, without saying so, AEDPA replaced the Brecht standard of “‘actual prejudice,’” 507 U. S., at 637 (quoting United States v. Lane, 474 U. S. 438, 449 (1986)), with the more liberal AEDPA/Chapman standard which requires only that the state court’s harmless-beyond-a-reasonable-doubt determination be unreasonable.

It is not self-evident that the AEDPA/Chapman standard is, in fact, “more liberal.” “Unreasonable” is a strong word, and a state court has to be far out of line before its decision is not merely incorrect but unreasonable. Justice Scalia's point, however, is valid that it is implausible that Congress intended to remove the Brecht hurdle. The next statement, though, is pure dictum: “That said, it certainly makes no sense to require formal application of both tests (AEDPA/Chapman and Brecht) when the latter obviously subsumes the former.”

No, it is not obvious. Whether the petitioner can clear the Brecht hurdle and whether the state court's Chapman analysis was beyond incorrect into unreasonable are distinct inquiries, just as the state court's reasonableness on a question of law and the limit on retroactivity of Teague v. Lane, 489 U. S. 288 (1989) are distinct inquiries. See Horn v. Banks, 536 U. S. 266, 272 (2002).

Does it matter in practice? Probably not.  The standards involved here are not mathematically precise. They are vague and subjective.  Most judges in most cases would probably decide the same way under either standard. One inquiry is better than two at achieving one of AEDPA's goals, to speed up the process. A dual inquiry on harmless error is probably not appreciable better than one in achieving the other main goal, stopping the erroneous overturning of valid state-court judgments. Whatever its theoretical impurity, it may be better in the long run to just go along with the Fry dictum.