Abstract: Much criminal law theory attempts to formulate master principles of criminal law: harm principles, autonomy principles, fair labelling principles, and so on. But what are the criteria of success for such principles? Surprisingly, this question has received scant attention. Responding to recent discussions by Anthony Duff,Victor Tadros, and James Edwards, I set out three approaches to evaluating master principles: the traditional, intuitive approach; the 'all-at-once' proportionality approach; and the results-oriented policy approach. I argue that each approach faces considerable difficulties.