The success of captive-breeding and release programmes is often compromised by predation of released individuals, which are naïve about predators. Pre-release behavioural preparation of release candidates in the form of anti-predator training has been attempted infrequently, usually using models of predators, but success was most often measured in terms of improved behavioural responses rather than survival to breeding age after release. Here we report that post-release survival of captive-reared houbara bustards (Chlamydotis [undulata] macqueenii) was improved through exposure to a live predator before release: a result with possible applications for a wide range of species currently the focus of reintroduction projects. We also show that rearing houbara with minimal human contact and training with a model of a predator had no effect on post-release survival. Moreover, neither pre-release behavioural responses to a model predator nor the degree of tonic immobility were predictors of post-release survival.