Despite strong selection on young and adult prey to avoid predators, we often find obvious differences in anti-predator behaviour between them. By examining the ontogeny of antipredator behaviour, we can gain insight into how young animals come to classify predators, knowledge crucial for their survival. In this thesis, I investigated how and when young meerkats (Suricata suricatta) develop the ability to produce alarm calls with an adult-like structure, how they come to use these alarm calls in appropriate contexts, and how they respond to the alarm calls of other group members. Meerkats, which are cooperatively breeding mongooses living under a high predation pressure, have evolved a sophisticated alarm-call system consisting of calls which simultaneously encode information about specific predator types and the level of urgency, and calls not distinctively related to specific predator types. This system is therefore highly suitable to address questions regarding vocal development of alarm calls. Since captive environments often lack many of the natural predators that wild animals experience, research on captive animals can also offer additional insight into how experience with predators may influence a species’ capabilities of responding to them. I collected behavioural observations of naturally occurring predator events, and conducted playback and manipulation experiments at different stages during juvenile development in wild and captive populations. Although young meerkats were able to respond correctly to alarm calls within three months, their probability of doing so increased as they grew older. Young were also likely to gather cues from other group members by looking towards them or running to them. These results suggest that experience is needed to perfect the alarm-call responses of young. However, young showed correct responses to alarm calls signalling predators closeby at an earlier age than to those signalling predators far away, indicating that responses may also be adapted to the level of risk posed by different situations. Correct responses were not, however, contingent upon the particular predator approaching. In terms of call production, young were less likely to utter alarm calls than were adults, but also less likely to look out for predators. Since alarm calls were more likely to be given by vigilant young, the increase in alertness with age might be responsible for an age-related increase in alarm calling. Nevertheless, alarm calls which are not related to specific predator types were produced much earlier than predator type specific calls, indicating that some learning may also be involved. Experience also seems necessary to restrict alarm calling to predators belonging to particular classes. In contrast, although the alarm calls of young underwent slight modification during development, changes which are likely to reflect physical maturation, they were more or less structurally indistinguishable from those of adult calls. All alarm calls that have been documented in the wild also occurred in captivity. The acoustic structure, however, differed slightly from that observed in the wild, but may only reflect differences in arousal. Without experience of odours from predators, captive-born meerkats distinguished between faeces of potential predators and non-predators, similar to that of wild individuals. Together, these findings show that young animals come to classify predators through a mixture of innately recognised features and gradual modification as a result of experience,and provide an important contribution to the small existing literature on predator avoidanceontogeny.