Sex pheromone communication in moths is a well investigated case of mate-finding by chemical signals, but the evolutionary causes of the great complexity and diversity of these signals are still not generally agreed on. In the present paper, I argue that there is no reason to dismiss species recognition as a possible cause of evolutionary change in moth sex pheromones. Admittedly, selection for species recognition cannot explain all of the diversity in sex pheromones and the data supporting this contention are weak, but the alternative causes suggested, invoking mate choice between conspecifics as the mechanism of sexual selection, has so far no empirical support. Finding and analysing genes responsible for mate choice is important to corroborate any theory of sexual selection and speciation. In this respect genetic dissection of moth pheromone communication has provided important progress. Mendelian genes controlling differences in mate choice and in the production of mate recognition signals have been found. Polymorphic pheromone systems give the population biologists unique possibilities to study mate choice and selection at the genotype level in nature.