Abstract Background The currently recognized species richness of South American salamanders is surprisingly low compared to North and Central America. In part, this low richness may be due to the salamanders being a recent arrival to South America. Additionally, the number of South American salamander species may be underestimated because of cryptic diversity. The aims of our present study were to infer evolutionary relationships, lineage diversity, and timing of divergence of the South American Bolitoglossa using mitochondrial and nuclear sequence data from specimens primarily from localities in the Andes and upper Amazon Basin. We also estimated time of colonization of South America to test whether it is consistent with arrival via the Panamanian Isthmus, or land bridge connection, at its traditionally assumed age of 3 million years. Results Divergence time estimates suggest that Bolitoglossa arrived in South America from Central America by at least the Early Miocene, ca. 23.6 MYA (95% HPD 15.9-30.3 MYA), and subsequently diversified. South American salamanders of the genus Bolitoglossa show strong phylogeographic structure at fine geographic scales and deep divergences at the mitochondrial gene cytochrome b (Cytb) and high diversity at the nuclear recombination activating gene-1 (Rag1). Species often contain multiple genetically divergent lineages that are occasionally geographically overlapping. Single specimens from two southeastern localities in Ecuador are sister to the equatoriana-peruviana clade and genetically distinct from all other species investigated to date. Another single exemplar from the Andes of northwestern Ecuador is highly divergent from all other specimens and is sister to all newly studied samples. Nevertheless, all sampled species of South American Bolitoglossa are members of a single clade that is one of several constituting the subgenus Eladinea, one of seven subgenera in this large genus. Conclusions The ancestors of South American salamanders likely arrived at least by the Early Miocene, well before the completion of the Late Pliocene Panamanian land bridge (widely accepted as ca. 3 MYA). This date is in agreement with recent, controversial, arguments that an older, perhaps short-lived, land connection may have existed between South America and present-day Panama 23–25 MYA. Since its arrival in South America, Bolitoglossa has diversified more extensively than previously presumed and currently includes several cryptic species within a relatively small geographic area. Rather than two upper Amazonian species currently recorded for this region, we propose that at least eight should be recognized, although these additional lineages remain to be formally described.