Dittrich et al. (Proc. R. Soc. Lond. B 251, 195 (1993)) recently proposed that the existence of hoverfly species which are `imperfect' mimics of wasps is a consequence of a constraint on the perceptual or learning mechanisms of birds. Here we put forward a new theory of the evolution of `imperfect mimics'. We propose that the apparent imperfection should be regarded as either true ambiguity, in which the vertebrate predator has difficulty in interpreting the meaning of two equally probable messages, or as the presentation of a sign in an incongruous context, which also interferes with the interpretation of the meaning of the message. In either case, the lengthening of the perceptual process will give the prey opportunity to escape. We term this satyric mimicry in reference to the ambiguity or paradox involved. We suggest explanations for the experimental results of Dittrich et al., and briefly discuss the importance of satyric mimicry in the evolution of Batesian mimicry.