The role of sexual selection in speciation is investigated, addressing two main issues. First, how do sexually selected traits become species recognition traits? Theory and empirical evidence suggest that female preferences often do not evolve as a correlated response to evolution of male traits. This implies that, contrary to runaway (Fisherian) models of sexual selection, premating isolation will not arise as an automatic side effect of divergence between populations in sexually selected traits. I evaluate premating isolating mechanisms in one group, the birds. In this group premating isolation is often a consequence of sexual imprinting, whereby young birds learn features of their parents and use these features in mate choice. Song, morphology and plumage are known recognition cues. I conclude that perhaps the main role for sexual selection in speciation is in generating differences between populations in traits. Sexual imprinting then leads to these traits being used as species recognition mechanisms. The second issue addressed in this paper is the role of sexual selection in adaptive radiation, again concentrating on birds. Ecological differences between species include large differences in size, which may in themselves be sufficient for species recognition, and differences in habitat, which seem to evolve frequently and at all stages of an adaptive radiation. Differences in habitat often cause song and plumage patterns to evolve as a result of sexual selection for efficient communication. Therefore sexual selection is likely to have an important role in generating premating isolating mechanisms throughout an adaptive radiation. It is also possible that sexual selection, by creating more allopatric species, creates more opportunity for ecological divergence to occur. The limited available evidence does not support this idea. A role for sexual selection in accelerating ecological diversification has yet to be demonstrated.