This Article explores whether and when rules of customary international law (CIL) can be expected to be efficient. Customary rules are often regarded as desirable because in certain circumstances, they promote the welfare of the group in which they arise. Unless these circumstances apply among states, the efficiency arguments for the legalization of customary norms do not apply. The Article takes as its central observation the divergent treatment of custom in domestic and international law. In international law, if a customary behavior of states can be identified, it is automatically elevated to the status of legal obligation without any independent examination of whether the custom is a good one. International custom is customary international law. This reification of custom is in marked contrast to the treatment of custom in private law. No one doubts that customary behaviors exist in various societal subgroups, but tort law does not assume that customs are normatively desirable, and does not automatically transform customs into legally binding obligations. Thus tort law does not take custom to dictate the standard of care; the fact-finder must independently determine whether the practice is efficient, though its customary status has some positive evidentiary value. Law and economics scholars have varied views about whether custom is presumptively efficient in the private law context. The most optimistic view holds that private custom will generally be welfare enhancing, and thus courts should give legal recognition to such practices. Yet even the optimistic view holds that efficient custom would only arise in certain circumstances: when there are thickly repeated dealings between members of an insular, homogenous group whose members play reciprocal roles. The Article takes these earmarks of efficient custom and examines whether they apply to international custom. It finds that much of international custom should not be expected to be efficient even in the most optimistic view of custom. Some areas of CIL, like diplomatic privileges, might satisfy efficient custom criteria. This suggests that, contrary to current practice, CIL should not be treated as one undifferentiated phenomenon. Rather, the standards for establishing a CIL norm should vary across different substantive contexts and different groups of states.