Even the most generalist parasites usually occur in only a subset of potential host species, a tendency which reflects overriding environmental constraints on their distributions in nature. The periodic shifting of these limitations represented by host-switches may have been an important process in the evolution of many host-parasite assemblages. To study such events, however, it must first be established where and when they have occurred. Past host-switches within a group of parasites are usually inferred from a comparison of the parasite phylogeny with that of the hosts. Congruence between the phylogenies is often attributed to a history of association by descent with cospeciation, and incongruence to host-switching or extinction in ‘duplicated’ parasite lineages (which diverged without a corresponding branching of the host tree). The inference of host-switching from incongrucnt patterns is discussed. Difficulties arise because incongruence can frequently be explained by different combinations of biologically distinct events whose relative probabilities are uncertain. Also, the models of host parasite relationships implicit in historical reconstructions may often not allow for plausible sources of incongruence other than host-switching or duplication/extinction, or for the possibility that colonization could, in some circumstances, be disguised by ‘false’ congruence.