Intraguild predation (IGP) is a commonly recognized mechanism influencing the community structure of predators, but the complex interactions are notoriously difficult to disentangle. The mesopredator suppression hypothesis predicts that a superpredator may either simultaneously repress two mesopredators, restrain the dominant one and thereby release the subdominant mesopredator, or elicit different responses by both mesopredators. We show the outcome arising from such conditions in a three-level predator assemblage (Eurasian eagle owl Bubo bubo L., northern goshawk Accipiter gentilis L. and common buzzard Buteo buteo L.) studied over 25 years. In the second half of the study period, the eagle owl re-colonized the study area, thereby providing a natural experiment of superpredator introduction. We combined this set-up with detailed GIS analysis of habitat use and a field experiment simulating intrusion by the superpredator into territories of the subdominant mesopredator, the buzzard. Although population trends were positive for all three species in the assemblage, the proportion of failed breeding attempts increased significantly in both mesopredators after the superpredator re-colonized the area. We predicted that superpredator-induced niche shifts in the dominant mesopredator may facilitate mesopredator coexistence in superpredator-free refugia. We found significant changes in nesting habitat choice in goshawk, but not in buzzard. Since competition for enemy-free refugia and the rapid increase in population density may have constrained niche shifts of the subdominant mesopredator, we further predicted behavioural changes in response to the superpredator. The field experiment indeed showed a significant increase in aggressive response of buzzards towards eagle owl territory intrusion over the course of 10 years, probably due to phenotypic plasticity in the response towards superpredation risk. Overall, our results show that intraguild predation can be a powerful force of behavioural change, simultaneously influencing habitat use and aggressiveness in predator communities. These changes might help to buffer mesopredator populations against the negative effects of intraguild predation.