What is the proof standard for applying preventive criminal sanctions? This is an open question in various legal systems. Some authors suggest that we can answer it by using decision theory. On this approach, the proof standard is conceptualized as a probabilistic threshold: a preventive sanction can only be imposed if it is sufficiently probable that a person will commit a crime in the future. According to decision-theorists, how high this probability of a future crime should be, can be determined by means of a utilitarian calculus. However, such a decision-theoretic analysis requires wrestling with a number of difficult questions. This article surveys these questions and explores some avenues for answering them. It does so by considering a standard of proof for a fictional preventive sanction and offering a decision-theoretic justification for that standard.