Despite decades of interest, the adaptive significance of the extraordinary diversity in the design of animal signals remains elusive. It is suggested that signal design consists of two components: 'strategic design' and 'efficacy'. Strategic design is concerned with how a signal is constructed by natural selection to provide the information necessary to make a receiver respond (e.g. by being good at displaying underlying quality), whilst efficacy is concerned with how a signal is designed to get that information across to the receiver (e.g. by being easily measured). It is argued that an important but neglected evolutionary force on animal signals is therefore the psychology of the signal receiver, and that three aspects of receiver psychology (what a receiver finds easy to detect, easy to discriminate and easy to remember) constitute powerful selective forces in signal design. Greatest emphasis is given to memorability because this has been least considered by previous authors. It is argued that learning and memory are involved in a wide range of signals, and numerous hypotheses as to how signals may be adapted to be more memorable to receivers are suggested. The relationship of this analysis to earlier attempts at understanding signals is explored, particularly with reference to the concepts of honesty, manipulation and mind-reading. © 1991 The Association for the Study of Animal Behaviour.