Aggression directed by 53 potential host species towards a dummy of the parasitic common cuckoo, Cuculus canorus, was tested in relation to their breeding habitat, their suitability as a host and whether they were breeding in sympatry or not with the cuckoo. Host habitats were divided into three categories: (1) always breeding near trees, (2) some populations breeding near trees, others in open areas, and (3) always breeding in open areas. Each species was also placed in one of five categories according to their suitability as a cuckoo host. Strong support was found for predictions derived from the 'spatial habitat structure hypothesis', which argues that common cuckoos only breed in areas where they have access to vantage points in trees. Thus, species which have some populations breeding near trees and others breeding further from trees have a different cuckoo-host population dynamics than species that always breed near trees, or always breed in open areas. Aggression levels were highest among species regarded as being always suitable as hosts, and species which always breed near trees. However, populations breeding in sympatry with the cuckoo were more aggressive than allopatric populations, indicating the plasticity of aggressive behaviour. Adaptive behaviour in cuckoo hosts can be predicted from the 'spatial habitat structure hypothesis'.