In many group-living species, individuals are required to flexibly modify their communicative behaviour in response to current social challenges. To unravel whether sociality and communication systems co-evolve, research efforts have often targeted the links between social organisation and communicative repertoires. However, it is still unclear which social or interactional factors directly predict communicative complexity. To address this issue, we studied wild and zoo-housed immature orangutans of two species to assess the impact of the socio-ecological setting on the production of non-vocal signal repertoires. Specifically, we compared repertoire size, dyadic repertoire similarity, and number of social goals (i.e. observer’s estimate of the signaller’s intended interaction outcome) for communicative interactions with mothers versus other conspecifics, controlling for critical individual and environmental factors. In this small sample of immature orangutans, wild-captive contrasts were statistically significant only for other-directed repertoires, but not for mother-directed repertoires, and not for the number of social goals that immatures communicated towards. While the repertoires of individuals living in the same research setting were more similar than those living in contrasting settings, this difference was most pronounced for other-directed repertoires of the less socially tolerant orangutan species. These results suggest that the boosted interactional opportunities in captivity rather than mere differences in environmental affordances or communicative needs drive the wild-captive contrast in orangutan communicative repertoires. Overall, this fine-grained analysis of repertoires further underscores that not only a species’ social organisation but also the targeted audience may have a profound impact on communicative behaviour.Significance statementNavigating a dynamic social environment often requires flexible signal use. While it has repeatedly been shown that the social organisation and structure of species predict the complexity of their communication systems, the mechanisms underlying these relationships are largely unknown. Because targeted studies to assess this issue in great apes are difficult, we take an alternative approach here: we compare the same species living in the wild and in artificial habitats in captivity. This contrast allows a direct test of how repertoires respond to the relevant difference in socio-ecological conditions. Our results show that the diversity of interaction partners (i.e. social opportunities), but not the diversity of social goals (i.e. possible interaction outcomes) or the broader physical opportunities (i.e. safe ground use), predict the size and consistency of wild and captive signalling repertoires.