Résumé: AimThe delineation of regions is a critical procedure in biogeography, but there is still no consensus about the best approach. Traditionally, a compositional dissimilarity index and a clustering algorithm are used to partition locations into regions. However, the choice of index and algorithm may have a profound impact on the final result, particularly when locations display different levels of species richness and when they are nested within each other. Our objective was to estimate the influence of species nestedness among locations on the delineation of biogeographical regions. LocationAs a case study, we used coral reef fishes (families Chaetodontidae, Pomacentridae and Labridae) from the Indo-Pacific, where a large richness gradient extends, often as a series of nested assemblages, from the species-rich Indo-Australian Archipelago (Coral Triangle) to species-poor peripheral locations. MethodsWe used the turnover and nestedness components of the SOrensen and Jaccard dissimilarity indices to estimate the effect of nestedness on the delineation of biogeographical regions. In addition, we compared the results with those obtained using a parsimony analysis of endemicity (PAE). ResultsLow Mantel correlation values revealed that the PAE method assembled locations in a very different way than methods based on dissimilarity indices for Indo-Pacific coral reef fishes. We also found that nestedness mattered when delineating biogeographical units because, for both the SOrensen and the Jaccard indices, reef fish assemblages were grouped differently depending on whether we used the turnover component of each index or the complete index, including the nestedness component. The turnover component ignored variation in species richness attributable to differences in habitat area between locations, and permitted a delineation based solely on species replacement. Main conclusionsWe demonstrate that the choice of the component used to measure dissimilarity between species assemblages is critical, because it may strongly influence regional delineations, at least for Indo-Pacific coral reef fishes. We conclude that the two components of the dissimilarity indices can reveal complementary insights into the role that history may have played in shaping extant patterns of biodiversity.