In Williams v. Taylor and Ramdass v. Angelone, the United States Supreme Court confronted one of the core provisions of the Antiterrorism and Effective Death Penalty Act of 1996 and confirmed what some had hoped and others had feared: A federal court may no longer grant habeas relief merely because it concludes that a state court erroneously applied federal law when it rejected a state prisoner’s federal constitutional claim. Instead, a federal court must deny habeas relief to a state prisoner whose federal constitutional rights have been violated and whose requests for relief have been erroneously denied by a state court, so long as the state court’s erroneous application of law to fact was objectively reasonable. It is not clear, however, how courts should distinguish between reasonably and unreasonably erroneous applications of federal law. After briefly describing the evolution of the federal habeas standard of review for mixed questions of law and fact, this article points out the ways in which several well-known standards of review fail to illuminate the manner in which courts are to apply the new “unreasonably erroneous” standard. The article then discusses the standard’s implicit rejection of two theories of adjudication—deterministic formalism and indeterministic skepticism; examines the contributions of conventionalism; and illustrates the difficulties one faces when trying to frame a theory of adjudication that will enable federal courts to apply the “unreasonably erroneous” standard in a persuasively principled manner in all cases. Finally, the article proposes three analytic touchstones that, in the absence of an overarching theory of adjudication, can help federal courts determine the likelihood that state courts’ rulings should be deemed objectively unreasonable.