At broad scales in the Amazon, it is often hypothesized that species distributions are limited by geographical barriers, such as large rivers (river-barrier hypothesis). This hypothesis has been used to explain the spatial-distribution limits of species and to indicate endemism areas for several phylogenetic lineages. We tested the ability of the river-barrier hypothesis to explain patterns of species diversity and spatial-distribution limits for 1952 easily-detected species in 14 taxonomic groups that occur around the Madeira River, and our results indicate that the hypothesis that the Madeira River is the border between endemism areas and explains much of the diversity found in the region is inappropriate for >99% of species. This indicates that alternative hypotheses should be proposed to explain the limits of distributions of species around the Madeira River, as well as a revision of the criteria that are used to determine species-endemism areas.