We inferred the phylogenetic relationships among members of the Poecilia sphenops species complex to resolve the colonization process and radiation of this group in Central America. We analyzed 2550 base pairs (bp) of mitocondrial DNA (mtDNA), including ATP synthase 6 and 8, cytochrome oxidase subunit I and NADH dehydrogenase subunit 2 genes, and 906 bp of the nuclear S7 ribosomal protein of 86 ingroup individuals from 61 localities spanning most of its distribution from Mexico to Panama. Our mitochondrial data rendered a well-supported phylogeny for the P. sphenops complex that differed with the nuclear data set topology, which did not recover the monophyly of the P. mexicana mitochondrial lineage. Coalescent-based simulations tests indicated that, although hybridization cannot be completely ruled out, this incongruence is most likely due to incomplete lineage sorting in this group, which also showed the widest geographic distribution. A single colonization event of Central America from South America was estimated to have occurred between the early Paleocene and Oligocene (53 – 22 million years ago). Subsequently, two largely differentiated evolutionary lineages diverged around the Early Oligocene-Miocene (38 – 13 million years ago), which are considered two separate species complexes: P. sphenops and P. mexicana, which can also be distinguished by their tricuspid and unicuspid inner jaw teeth, respectively. Ultimately, within lineage diversification occurred mainly during the Miocene (22 – 5 million years ago). All major cladogenetic events predated the final closure of the Isthmus of Panama. The allopatric distribution of lineages together with the long basal internodes suggest that vicariance and long term isolations could be the main evolutionary forces promoting radiation in this group, although dispersal through water barriers might also have occurred. Lastly, our results suggest the need to review the current species distribution and taxonomy of the P. sphenops complex sensu lato.