Captive breeding of threatened species, for release to the wild, is critical for conservation. This strategy, however, risks producing captive-raised animals with traits poorly suited to the wild. We describe the first study to characterise accumulated consequences of long-term captive breeding on behaviour, by following the release of Tasmanian devils to the wild. We test the impact of prolonged captive breeding on the probability that captive-raised animals are fatally struck by vehicles. Multiple generations of captive breeding increased the probability that individuals were fatally struck, a pattern that could not be explained by other confounding factors (e.g. age or release site). Our results imply that long-term captive breeding programs may produce animals that are naïve to the risks of the post-release environment. Our analyses have already induced changes in management policy of this endangered species, and serve as model of productive synergy between ecological monitoring and conservation strategy.