Many rules of evidence employ balancing tests. The best known is Rule 403, requiring a judge to compare the probative value of evidence against (for example) its prejudicial effect. The problem of metacognition arises when a trial judge is making a decision in a specific case. The required balancing involves more than knowing a rule and the import of the evidence: it engages the judge’s beliefs about how jurors’ cognitive and emotional processes work, predicting how they will respond to the evidence, and comparing that to how the judge thinks the evidence ought best to be responded to. For some categories of evidence, the rulemakers have performed the metacognitive balancing for the judge, deciding that because jurors likely will over-value the evidence, it is best to prevent the jury from hearing those types of evidence. Sometimes the rules aim to encourage socially desirable behavior. For example, to promote repairs of dangerous situations, the rules bar using the fact that repairs were made to prove negligence. A well-established body of research on the “hindsight bias” supports the law’s suspicion that jurors will use evidence of the accident and the repair to answer the question of whether the risk should have been recognized ex ante, and conclude that the failure to correct the situation earlier was negligent. Balancing tests make good psychological sense in theory. Much remains to be learned, however, about whether they “work” (that is, are efficient, lead to good results, encourage socially-desirable behavior) in practice.