Brown v. Davenport
The Supreme Court ruled that federal courts must clear two separate legal hurdles — not just one — before overturning a state criminal conviction through federal habeas review.
The decision resolves a split among federal appeals courts and reinforces Congress's 1996 law tightly limiting when federal courts can second-guess state courts, making it harder for state prisoners to win federal relief.
How it got here: Michigan state courts found Davenport's shackling error harmless; a federal district court denied habeas relief under AEDPA; a divided Sixth Circuit reversed applying only Brecht; the Supreme Court granted certiorari to resolve a circuit split.
The Case in Depth
What happened
Ervine Davenport was convicted of first-degree murder in Michigan after strangling a woman he was driving home. During trial he was partially shackled — a constitutional violation under the Supreme Court's Deck v. Missouri ruling — but Michigan courts found the error harmless beyond a reasonable doubt after hearing from all twelve jurors, who testified the restraints had not affected their verdict. The evidence of guilt was described by courts as overwhelming.
The question before the Court
When a state court has already ruled on a prisoner's claim, must a federal court apply both the Brecht standard and Congress's AEDPA standard before granting habeas relief, or is satisfying Brecht alone enough?
The Court's answer
Yes — both tests must be cleared. When a state court has already decided a prisoner's claim on the merits, a federal court cannot grant habeas relief unless the prisoner satisfies both the Brecht standard and the requirements Congress set in AEDPA.
Brecht requires a prisoner to show that a trial error had a "substantial and injurious effect" on the verdict — but that is only a necessary condition, not a sufficient one. Congress in AEDPA separately commanded that federal courts "shall not" grant relief unless the state court's decision was unreasonable under clearly established Supreme Court law. The two tests ask different questions and draw on different legal materials: Brecht asks whether the habeas court itself doubts the verdict; AEDPA asks whether any reasonable jurist could have agreed with the state court. Applied here, even assuming Davenport satisfied Brecht, Michigan's courts reasonably found the shackling harmless — the evidence of guilt was overwhelming and all twelve jurors testified the shackles had not influenced their deliberations.
Curious how the Court got there? See the step-by-step legal reasoning →
Why it matters
State prisoners challenging their convictions in federal court now face a confirmed two-step process: they must show actual prejudice under Brecht and separately show that the state court acted unreasonably under AEDPA. Courts and prisoners that had followed the Sixth Circuit's one-step approach must now do both analyses, and more petitions are likely to fail at the AEDPA stage even when a federal judge personally doubts the verdict.
What changes now
The Sixth Circuit's order requiring Michigan to retry or release Davenport is reversed. He remains convicted of first-degree murder, and the federal courts' denial of habeas relief stands. Going forward, every federal court handling a state prisoner's habeas petition — where the state court ruled on the merits — must conduct both a Brecht analysis and an AEDPA analysis before granting relief, regardless of how clearly the petitioner clears the Brecht hurdle.
What this does not decide
The Court assumed without deciding that Davenport actually satisfied Brecht; it ruled only that he could not satisfy AEDPA. The ruling also does not affect cases where a state court never ruled on the merits of the prisoner's claim, in which AEDPA does not apply and Brecht governs alone.
Concurrences and dissents
Dissent — Justice Kagan
Justice Kagan argued that two earlier unanimous Supreme Court decisions — Fry v. Pliler and Davis v. Ayala — already clearly resolved this question: because Brecht is harder for prisoners to satisfy than AEDPA/Chapman, satisfying Brecht necessarily means satisfying AEDPA, so requiring formal application of both tests is pointless make-work. She accused the majority of distorting those precedents beyond recognition and predicted the dual-test requirement would never change a habeas outcome, since no reported case has ever found Brecht satisfied but AEDPA not satisfied.
How the Court got there
The legal reasoning, step by step
- The Court began with AEDPA's text: Congress commanded that a federal court 'shall not' grant habeas relief when a state court has adjudicated a claim on the merits unless that decision was contrary to, or an unreasonable application of, clearly established Supreme Court law, or based on an unreasonable factual finding. That is a mandatory statutory command federal courts must follow.
- Separately, the Court's 1993 Brecht decision — a judge-made rule rooted in federal courts' traditional equitable discretion — requires a state prisoner to show that a constitutional trial error had a 'substantial and injurious effect or influence' on the verdict. AEDPA left this equitable tradition intact, so Brecht continues to apply alongside AEDPA.
- The two tests ask analytically distinct questions. AEDPA asks whether every reasonable jurist would agree the state court's harmlessness finding was wrong; Brecht asks only whether the habeas court itself harbors 'grave doubt.' AEDPA also limits courts to Supreme Court precedents; Brecht permits reliance on the full body of law, including circuit court decisions. A petitioner could satisfy Brecht using circuit precedent that does not qualify as 'clearly established' Supreme Court law under AEDPA.
- The Court rejected Davenport's argument that Brecht logically 'subsumes' AEDPA — meaning a Brecht win automatically delivers an AEDPA win. It read prior decisions Fry v. Pliler and Davis v. Ayala as not resolving this question; those cases' passing language about Brecht subsuming AEDPA addressed the scenario where a petitioner fails Brecht (making AEDPA analysis unnecessary), not the scenario where a petitioner passes Brecht.
- Applying AEDPA to Davenport's shackling claim, the Court found no unreasonable state-court ruling. The key precedent Davenport cited, Holbrook v. Flynn, addressed whether a trial error occurred at all — not whether an error was harmless — and its skepticism about prospective juror speculation did not bar state courts from relying on post-trial testimony from actual jurors. Davenport's alternative theory (that shackling influenced the degree of murder conviction) was never raised in state court and was therefore forfeited.
Doctrinal impact
Cases affected by this decision
Distinguishes Fry v. Pliler (551 U.S. 112)
Clarified that Fry's 'subsumes' language applied only when a petitioner fails Brecht, not when he passes it.
Distinguishes Davis v. Ayala (576 U.S. 257)
Clarified that Ayala's restatement of Fry similarly did not resolve whether passing Brecht is sufficient for relief.
Reaffirms Brecht v. Abrahamson (507 U.S. 619)
Reaffirmed that Brecht still applies in federal habeas cases, but as a necessary condition alongside AEDPA, not a standalone test.