Empirical studies show that courts frequently cite cases that the parties did not cite during briefing and oral arguments—endogenous cases. This Article shows the cognitive and rational dangers of endogenous cases and presents an empirical study of their use. I contend that judges should avoid using endogenous cases in their reasoning and opinions. This Article’s first significant contribution is to provide the first exhaustive treatment in the American legal literature of the rational bases upon which defeasible legal deductions and legal analogies may be built and the critical questions or defeaters that can weaken or bring them down. As far as we know, Aristotle was the first Westerner to formalize logic, about 2,500 years ago. He recognized immediately, however, that the valid de-ductive forms of reasoning he described do not take us far in reasoning about human affairs. Contemporary law scholars nevertheless talk about rule-based and deductive reasoning as if it exists in the law. But all legal arguments, even apparently deductive ones, are subject to defeat by counterargument—they are defeasible. This Article shows that legal analogies are in a sense more defeasible, and thus more complex, than legal deductions, and it suggests that legal policy arguments are more defeasible still.

But what happens when judges build these arguments around cases that the parties have not cited—endogenous cases? Studies suggest that judges do so about half the time. The theory of defeasible arguments suggests judges should be reluctant to do so, and they should be most reluctant to do so with legal analogies and policy arguments. But until now, no study has examined how judges use en-dogenous cases. This Article’s second significant contribution is an empirical study of cases judges cited in a random sample of federal district court opinions, identifying where they found them and how they used them. It also identifies a hi-erarchy of badness among endogenous case uses, warning judges away from the most dangerous. It concludes that judges should avoid endogenous cases or, at a minimum, permit the parties to argue the cases before their application. Though the opinions studied here were from federal trial courts, the principles extend to any tribunal that uses opinions from previous cases to guide its decision-making.