The high biodiversity of tropical marine hotspots has long intrigued evolutionary biologists and biogeographers. The genus Haemulon (grunts) is one of the most important (numerically, ecologically, and economically) reef fish groups in the New World and an excellent candidate to test hypotheses of speciation and diversity generation in the Greater Caribbean, the richest Atlantic biodiversity hotspot, as well as the eastern Pacific. To elucidate the phylogenetic relationships among the species of Haemulon, we obtained a combined total of 2639 base pairs from two mitochondrial genes (cytochrome b and cytochrome oxidase I), and two nuclear genes (TMO-4C4 and RAG2) from all nominal species. Parsimony, Maximum likelihood, and Bayesian analyses resulted in a well-resolved phylogeny with almost identical topologies. Previous phylogenetic hypotheses based on adult morphology, such as the close relationship among H. aurolineatum, H. boschmae, and H. striatum were not supported, whereas others using developmental characters, such as the relationship between H. plumieri and H. sciurus, were confirmed. Our data also indicate that the populations of the nominal H. steindachneri from the two sides of the Isthmus of Panama are genetically divergent at the species level in each ocean, and that the boga, Inermia vittata (family Inermiidae), belongs in the genus Haemulon. This evidence implies that there are 21 valid species of Haemulon, two more than previously recognized. The Amazon barrier and the Isthmus of Panama seem to have played roles in allopatric speciation of Haemulon. However, the majority of sister species pairs have completely overlapping distributions, indicating that vicariance is not the only process driving speciation in this genus. We conclude that both vicariance between biogeographic provinces, and ecological mechanisms of speciation within provinces contribute to species richness in the genus Haemulon.