Predicting the effects of changes in cover of live coral on assemblages of reef fishes will require that relationships between functional or taxonomic groups of fishes and their habitats be considered in the context of spatial scale and intensity of environmental disturbances. Beyond a general expectation that species richness and abundance of fish will rise with initial increases in coral cover from very low values, we know comparatively little regarding how these attributes and species composition of fish assemblages vary across a wide range in cover of live coral. Yet knowledge of such functions will provide great insight into expected responses of the fish assemblage to fluctuations in cover of coral or other key habitats, including the degree of resistance to changes in habitat and the possibility of threshold responses. We explored this issue for an assemblage of reef fishes in lagoons at Moorea, French Polynesia. We conducted surveys in 500 m2 plots that represented a range in cover of live coral (1 to 47 percent) and examined relationships between coral cover and species richness, taxonomic composition and population abundances. A similarity value was calculated between the assemblage of fish at each local site and that of the site with the highest cover of live coral. When plotted against the cover of live coral, the similarity function rapidly rose to an asymptotic value with initial increases in cover of live coral from low values. This nonlinear relationship suggested that assemblages of reef fish were broadly similar across a wide range of coral cover, and that responses to variation in coral cover would be difficult to detect when coral was abundant, but the same proportional decline in cover could produce sizeable changes in the assemblage when coral was rare.