As biodiversity hotspots, montane regions have been a focus of research to understand the divergence process. Like their oceanic counterparts, the diversity of the “sky islands” might be ascribed to geographic isolation of mountaintops. However, because the sky islands, and especially those in northern latitudes, are subject to extreme climatic events such as the glacial cycles that drove both altitudinal and geographical shifts in species’ distributions, the dynamic colonization process is also a possible factor driving divergence. Here we test these two hypotheses (i.e., isolation versus colonization) in a flightless montane grasshopper, Melanoplus oregonensis, which is a member of a diverse group that radiated across the Rocky Mountains of North America. Using approximate Bayesian computation (ABC) and spatially explicit simulations that account for spatial heterogeneity and temporal shifts in species distributions, we show that a colonization model of the sky islands from refugial populations provides a significantly better fit to the empirical genetic data than a model of the geographic isolation among sky islands. Moreover, support for the colonization model holds irrespective of whether the movement of individuals was modeled as a diffusion process or was informed by differences in habitat suitabilities across the landscape. With validation analyses to confirm the models provide a good fit to the data, as well as general power and quality analyses, the research not only adds to a growing body of work on the complex dynamics underlying montane biodiversity, but it also provides much needed evaluation of competing hypotheses based on explicit models of the divergence process, as opposed to inferences about diversification drivers from species diversity patterns.