There are two subspecies of the zebra finch, Taeniopygia guttata guttata and Taeniopygia guttata castanotis. T. g. guttata is found on the Lesser Sunda islands in Indonesia and the males differ from their Australian counterparts, T. g. castanotis, in having a thinner breast-band and grey chin and throat instead of the black and white throat bars. The songs of male guttata are longer and sung at a higher frequency than those of male castanotis. In contrast to the substantial differences between the two subspecies, there is little geographic variation with subspecies. In a recent aviary study of the social interactions and pair formation between members of captive colonies of guttata and castanotis, members of the two subspecies were observed to mate assortatively, i.e. guttata and castanotis did not form mixed pairs (Bohner et al. 1984). This raises the question of which cues ensure that the two subspecies are behaviourally isolated and hence mate assortatively. In song playback experiments, females of both subspecies discriminated between guttata and castanotis songs, preferring the songs of males of their own subspecies. In multiple mate choice tests and observations of the same individuals during pair formation in aviaries, male and female guttata and castanotis were found to prefer members of their own subspecies. However, guttata males that were painted to resemble castanotis males were preferred by castanotis females over unpainted guttata males whereas guttata females preferred the unpainted guttata males. In the aviary, the castanotis females paired with painted guttata males and guttata females paired with unpainted guttata males. These findings suggest that differences between the two subspecies in both song and breast-band size could play a role in mate choice and subspecies discrimination, thus leading to assortative mating between the two subspecies in capacity. To assess the importance of early rearing experience on the development of these visual and vocal differences between the two subspecies and its effect on the development of sexual preferences, guttata and castanotis that had been cross-fostered to the other subspecies were compared with those that has been normally raised by members of their own species. When cross-fostered to the other subspecies, castanotis and guttata males resembled their own subspecies in the macrostructural features of song which distinguish the two subspecies' songs. Hybrid males that were raised by one guttata and one castanotis parent have songs that are intermediate between those of guttata and those of castanotis males. These results suggest that rearing experience has little, if any, effect on the development of these macrostructural song differences between the two subspecies. In playback experiments, females preferred the songs of their foster-father's subspecies, irrespective of whether the songs were from males that had been cross-fostered or normally raised. This indicates that females use these macrostructural differences in song for subspecies discrimination and that female song preferences are learnt. Cross-fostered zebra finches resemble normally raised members of their own subspecies in size and plumage, and hybrids were intermediate. In multiple mate choice tests, females preferred normally raised birds of their own subspecies over those that had been cross-fostered and over those of the other subspecies. Since cross-fostered males do not appear to differ from normally raised birds of the same subspecies in plumage, size or song, these results suggest that females may discriminate between guttata and castanotis males on the basis of behavioural cues. Studies of multiple mate choice and pair formation showed that when both sexes were cross-fostered to the other subspecies mixed pairs (guttata-castanotis) were formed, suggesting that early experience with the foster-parents can have an influence on pair formation through sexual imprinting on the parents. However, when one sex is normally raised and the other is cross-fostered, the cross-fostered birds usually pair with members of their own subspecies. This indicates that the likelihood of pairing with the `wrong' subspecies is reduced when cross-fostered individuals interact with members of their own subspecies. This would provide an adaptive mechanism for maintaining behavioural isolation between the two subspecies. Comparing the mate choice during one-way and two-way interaction suggests that normally raised zebra finches choose more often than cross-fostered birds and that, when both sexes are cross-fostered, the prerogative lies with the female.