A major influence on the evolution of insect song patterns is thought to be their importance as species recognition signals. This predicts that female preference should be species specific, i.e. females should prefer the song of their own species but not discriminate among common differences between males of their own species. This could result either through selection for species recognition during speciation or coevolution. An alternative view is that female preference may have evolved by sexual selection, with female preferences functioning to allow discrimination between potential mates of the same species. Recent work on frog acoustics has suggested yet another possibility is that preference and male trait can be substantially mismatched. Most work in insect bioacoustics has assumed that the species recognition function predominates. Here I describe work in progress on the song of the bushcricket Ephippiger ephippiger. This species shows unusual variability in song pattern. Female preferences change between geographic variants of male song, implying coevolution has occurred. The geographic variation occurs largely independently of the presence of similar species. Examination of the structural components of song which influence preference has implied that subtle aspects may be involved. Female preference is strongly influenced by changes which occur commonly between males within natural populations. These observations suggest sexual selection has had a greater impact on the evolution of song preference than species recognition. I shall discuss if this might be true generally.