Abstract Aim We study the Neotropical poison frogs of the genus Dendrobates Wagler, 1830 in order to clarify their phylogenetic relationships and biogeographical history. The genus Dendrobates is an excellent taxon for examining patterns of Neotropical diversification as the four major species groups appear to correspond roughly to distinct geographical regions: (1) trans-Andean, (2) Andean foreland, (3) Brazilian Shield and (4) Guianan Shield/Central America. In order to test the agreement of five of the most prominent hypotheses of Amazonian diversification, phylogenetic patterns were examined for agreement with patterns predicted by these hypotheses. Location Central and South America Methods The phylogenetic relationships of the genus Dendrobates were examined from novel and existing (GenBank) sequences of four mitochondrial loci totalling c. 1400 bp from 40 specimens of 22 different species using maximum parsimony and Bayesian methods. Results were compared with traditional taxonomic arrangements by means of SH tests. Phylogenetic relationships and genetic distances were used to test the adequacy of various diversification hypotheses. Results Phylogenetic analyses support the restructuring of two species groups of Dendrobates and the creation of a new species group. Statistical tests of the traditional taxonomic arrangement indicate a significantly bad fit to the molecular data. This restructuring has important implications for the understanding of the historical biogeography of Dendrobates. Biogeographical patterns within this genus suggest that a complex interaction of biotic and abiotic factors since the Eocene have produced the diversity observed today. Main conclusions The current classification of Dendrobates into discrete species groups does not accurately reflect evolutionary history. Data presented here strongly support a monophyletic Brazilian Shield lineage whose members have previously been split among the quinquevittatus and tinctorius groups. Furthermore, previous attempts at elucidating the historical biogeography of this genus were compromised by incomplete sampling and conclusions drawn from a paraphyletic ingroup. Our findings demonstrate a role for numerous hypotheses of diversification (e.g. river, refuge, disturbance–vicariance) in the history of Dendrobates, supporting previous warnings about the dangers of over-simplification in the study of Neotropical diversification.