Abstract: In 2009, the Court of Appeals for the Third Circuit decided Berrier v. Simplicity, a tragic, but otherwise modest personal injury diversity case that was brought under Pennsylvania products liability law. The Third Circuit predicted that Pennsylvania would adopt the Restatement (Third) of Torts, and abandon what everyone (including several members of the Pennsylvania Supreme Court) considered an incomprehensible products liability jurisprudence that had evolved under Pennsylvania’s interpretation of the Restatement (Second). But for five years the Pennsylvania Supreme Court did not adopt the Restatement (Third), despite several opportunities to do so. Yet during those years, the Third Circuit refused to back off its prediction, again, despite several opportunities to do so. As a result a party bringing a products liability action governed by the substantive law of Pennsylvania in a Pennsylvania state court would try the case under § 402 A of the Restatement (Second) of Torts. But if that party brought the same action in a federal court sitting in diversity, the court would try the case under the Restatement (Third) of Torts. The differences between the two Restatements would result in opposite outcomes in many, if not most cases. If we know anything about Erie, it is that federal courts sitting in diversity are supposed to apply the same substantive law as their state counterparts—any other result would be abhorrent, fundamentally unfair, and, according to accepted (although not uncontroversial) Erie jurisprudence, unconstitutional. Yet this is exactly what happened as a result of the Third Circuit’s prediction in Berrier, and its failure (driven by the law of the circuit doctrine) to back off this prediction. The problem is not unique to the Third Circuit. Wrong Erie guesses happen with surprising frequency in virtually all circuits. At a time when diversity cases make up a significant part of the federal court docket, this raises significant federalism issues and threatens to skew the delicate balance between the states and the federal sovereign. Further, it results in unfair disparate treatment of similarly situated parties. This article uses the Berrier line of cases as a backdrop to explore important issues involving the proper role of federal courts sitting in diversity in predicting what state courts will do in the face of uncertain state law and how the federal courts should respond when it appears their predictions are wrong. I first examine the problem of wrong Erie guesses, using Berrier as an illuminating example. I then examine the Erie decision in detail, including the nature of diversity jurisdiction, how Erie came to be decided, and the question of the decision’s constitutional basis. Concluding that Erie is constitutionally compelled as a matter of structural federalism (the Erie doctrine protects the crucial balance of authority struck by the constitution between the federal government and the several states), I take the position that the Erie guess process involves serious constitutional and fairness issues that must be considered in fashioning an approach to how federal courts ascertain state law in diversity cases, and more important, how they should respond when it becomes apparent that they have made wrong Erie guesses. I posit that diversity courts must be activist in how they ascertain state law (as the Berrier court was), using all the judicial resources available. Erie’s mandate that the federal diversity courts speak with a state court voice compels this; basic fairness to the litigants requires no less (the litigants should not be saddled with a “handicapped” judiciary just because they are in a federal diversity court). But, given this activist approach, I conclude that the federal courts must be more nimble in correcting wrong Erie guesses to avoid perpetuating the very two-tiered system of justice Erie sought to eradicate. This approach will help minimize diversity jurisdiction’s “federalism- tinged insult” and will help ensure that federal diversity courts (to paraphrase Professor Glassman) do less mischief if not none at all.