Aim: To understand why and when areas of endemism (provinces) of the tropical Atlantic Ocean were formed, how they relate to each other, and what processes have contributed to faunal enrichment. Location: Atlantic Ocean. Methods: The distributions of 2605 species of reef fishes were compiled for 25 areas of the Atlantic and southern Africa. Maximum-parsimony and distance analyses were employed to investigate biogeographical relationships among those areas. A collection of 26 phylogenies of various Atlantic reef fish taxa was used to assess patterns of origin and diversification relative to evolutionary scenarios based on spatio-temporal sequences of species splitting produced by geological and palaeoceanographic events. We present data on faunal (species and genera) richness, endemism patterns, diversity buildup (i.e. speciation processes), and evaluate the operation of the main biogeographical barriers and/or filters. Results: Phylogenetic (proportion of sister species) and distributional (number of shared species) patterns are generally concordant with recognized biogeographical provinces in the Atlantic. The highly uneven distribution of species in certain genera appears to be related to their origin, with highest species richness in areas with the greatest phylogenetic depth. Diversity buildup in Atlantic reef fishes involved (1) diversification within each province, (2) isolation as a result of biogeographical barriers, and (3) stochastic accretion by means of dispersal between provinces. The timing of divergence events is not concordant among taxonomic groups. The three soft (non-terrestrial) inter-regional barriers (mid-Atlantic, Amazon, and Benguela) clearly act as 'filters' by restricting dispersal but at the same time allowing occasional crossings that apparently lead to the establishment of new populations and species. Fluctuations in the effectiveness of the filters, combined with ecological differences among provinces, apparently provide a mechanism for much of the recent diversification of reef fishes in the Atlantic. Main conclusions: Our data set indicates that both historical events (e.g. Tethys closure) and relatively recent dispersal (with or without further speciation) have had a strong influence on Atlantic tropical marine biodiversity and have contributed to the biogeographical patterns we observe today; however, examples of the latter process outnumber those of the former. ?? 2007 The Authors.