Changing environments, whether through natural or anthropogenic causes, can lead to the loss of some selective pressures (‘relaxed selection’) and possibly even the reinstatement of selective agents not encountered for many generations (‘reversed selection’). We examined the outcome of relaxed and reversed selection in the adaptive radiation of the threespine stickleback fish, Gasterostues aculeatus L., in which isolated populations encounter a variety of predation regimes. Oceanic sticklebacks, which represent the ancestral founders of the freshwater radiation, encounter many piscivorous fish. Derived, freshwater populations, on the other hand, vary with respect to the presence of predators. Some populations encounter native salmonids, whereas others have not experienced predation by large fish in thousands of generations (relax-selected populations). Some relax-selected populations have had sport fish, including rainbow trout, Oncorhynchus mykiss , introduced within the past several decades (reverse-selected). We examined the behavioural responses of sticklebacks from three populations of each type to simulated attacks by trout and birds to determine whether relaxed and reversed selection has led to divergence in behaviour, and whether this divergence was predator specific. Fish from trout-free populations showed weak responses to trout, as predicted, but these responses were similar to those of oceanic (ancestral) populations. Fish from populations that co-occur with trout, whether native or introduced, showed elevated antipredator responses, indicating that in freshwater, trout predation selects for enhanced antipredator responses, which can evolve extremely rapidly. Comparison of laboratory-reared and wild-caught individuals suggests a combination of learned and genetic components to this variation. Responses to a model bird flyover were weakly linked to predation environment, indicating that the loss of predation by trout may partially influence the evolution of responses to birds. Our results reject the hypothesis that the consistent presence of predatory birds has been sufficient to maintain responses to piscivorous fish under periods of relaxed selection. [ABSTRACT FROM AUTHOR]