The results of a study of starling vocal mimicry are summarised and discussed in relation to various possible functional explanations of mimicry. Only one, a hypothesis based on sexual selection, is not rejected outright and even this was seen to have problems. A dif ferent approach yielded a causal hypothesis based on mistakes made during song learning. This causal hypothesis is consistent with all known features of starling mimicry, and can readily be extended to other species, but is probably not consistent with any of the candidate functional hypotheses, including the one based on sexual selection. It is tentatively concluded that there is no additional functional explanation of starling vocal mimicry.