This article addresses the question of whether the relation of moral preference is transitive. I argue, following Larry Temkin and Stuart Rachels, that any ethical theory complex enough to be even minimally plausible allows us to generate intransitive sets of preferences. Even act utilitarianism cannot avoid this predicament unless we accept its least plausible version. We must reevaluate the assumption that an ethical theory must be transitive in order to be rational. This problem amounts to a foundational crisis in ethics. However, it has not been taken seriously for two reasons—the belief that the problem has limited scope; and the claim that arguments against transitivity are 'merely' Sorites arguments. This article responds to both of these objections. I also point out some connections between intransitivity and the debate surrounding skepticism about the moral significance of numbers.