- Aim We examined the relationship between host species richness and parasite species richness using simultaneously collected data on small mammals (Insectivora, Rodentia and Lagomorpha) and their flea parasites. Location The study used previously published data on small mammals and their fleas from 37 different regions. All the world's main geographical regions other than Australasia and Wallacea were represented in the study, i.e. neotropical, nearctic, palaearctic, oriental and afrotropical realms. Methods We controlled the data for the area sampled and sampling effort and then tested this relationship using both cross-region conventional analysis and the independent contrasts method (to control for the effects of biogeographic historical relationships among different regions). Brooks parsimony analysis was used to construct a region cladogram based on the presence/absence of a host species and host phylogeny. Results Both cross-region and independent contrasts analyses showed a positive correlation between host species richness and flea species richness. Conventional cross-region regression under- or overestimated fleas species richness in the majority of regions. Main conclusions When the regression derived by the independent contrasts method was mapped onto the original tip data space, points that deviated significantly from the regression originated from Kenya, Mississippi and southern California (lower than expected flea richness) and Chile, Idaho, south-western California and Kyrgyzstan (higher than expected flea richness). These deviations can be explained by the environmental mediation of host–flea relationships and by a degree of environmental variety in sampled areas.