Fueled by police reliance on offender databases and advances in crime scene recovery, a new type of prosecution has emerged in which the government’s case turns on a match statistic explaining the significance of a “cold hit” between the defendant’s DNA profile and the crime-scene evidence. Such cases are unique in that the strength of the match depends on evidence that is almost entirely quantifiable. Despite the growing number of these cases, the critical jurisprudential questions they raise about the proper role of probabilistic evidence, and courts’ routine misapprehension of match statistics, no framework—including a workable standard of proof—currently exists for determining sufficiency of the evidence in such a case. This Article is the first to interrogate the relationship between “reasonable doubt” and statistical certainty in the context of cold hit DNA matches. Examining the concepts of “actual belief” and “moral certainty” underlying the “reasonable doubt” test, I argue that astronomically high source probabilities, while fallible, are capable of meeting the standard for conviction. Nevertheless, the starkly numerical nature of “pure cold hit” evidence raises unique issues that require courts to apply a quantified threshold for sufficiency purposes. I suggest as a starting point—citing recent juror studies and the need for uniformity and systemic legitimacy—that the threshold should be no less favorable to the defendant than a 99.9% source probability.