In wide-ranging taxa with historically dynamic ranges, past allopatric isolation and range expansion can both influence the current structure of genetic diversity. Considering alternate historical scenarios involving expansion from either a single refugium or from multiple refugia can be useful in differentiating the effects of isolation and expansion. Here, we examined patterns of genetic variability in the trans-continentally distributed painted turtle (Chrysemys picta). We utilized an existing phylogeographic dataset for the mitochondrial control region and generated additional data from nine populations for the mitochondrial control region (n = 302) and for eleven nuclear microsatellite loci (n = 247). We created a present-day ecological niche model (ENM) for C. picta and hindcast this model to three reconstructions of historical climate to define three potential scenarios with one, two, or three refugia. Finally, we employed spatially-explicit coalescent simulations and an approximate Bayesian computation (ABC) framework to test which scenario best fit the observed genetic data. Simulations indicated that phylogeographic and multilocus population-level sampling both could differentiate among refugial scenarios, although inferences made using mitochondrial data were less accurate when a longer coalescence time was assumed. Furthermore, all empirical genetic datasets were most consistent with expansion from a single refugium based on ABC. Our results indicate a stronger role for post-glacial range expansion, rather than isolation in allopatric refugia followed by range expansion, in structuring diversity in this species. To distinguish among complex historical scenarios, we recommend explicitly modeling the effects of range expansion and evaluating alternate refugial scenarios for wide-ranging taxa.