For over 50 years, courts and scholars have tried to conceptualize fact-finding, and burdens of persuasion, in terms of the probability of facts given the evidence. The exercise has not produced a satisfying theory of fact-finding. The problem is reliance on probability. Fact-finding is not about probability. It’s about likelihood. The difference between these concepts is substantial. Where probability theories of fact-finding ask about the probability of the facts given the evidence, the proposed likelihood approach asks about the probability of the evidence given different assumptions about the facts. Where probability theories measure subjective beliefs, the likelihood approach measures the relative weight of evidence alone. Using the statistical properties of likelihoods, I show that every burden of persuasion in use today can be reduced to the same simple rule of likelihood reasoning. This likelihood theory of fact-finding closely mirrors the procedure of adversarial litigation, and solves all of the paradoxes, difficulties, and unacceptable implications that have long frustrated probability theories of fact-finding.