Questions about how shifting distributions contribute to species diversification remain virtually without answer, even though rapid climate change during the Pleistocene clearly impacted genetic variation within many species. One factor that has prevented this question from being adequately addressed is the lack of precision associated with estimates of species divergence made from a single genetic locus and without incorporating processes that are biologically important as populations diverge. Analysis of DNA sequences from multiple variable loci in a coalescent framework that (i) corrects for gene divergence pre-dating speciation, and (ii) derives divergence-time estimates without making a priori assumptions about the processes underlying patterns of incomplete lineage sorting between species (i.e. allows for the possibility of gene flow during speciation), is critical to overcoming the inherent logistical and analytical difficulties of inferring the timing and mode of speciation during the dynamic Pleistocene. Estimates of species divergence that ignore these processes, use single locus data, or do both can dramatically overestimate species divergence. For example, using a coalescent approach with data from six loci, the divergence between two species of montane Melanoplus grasshoppers is estimated at between 200,000 and 300,000 years before present, far more recently than divergence estimates made using single-locus data or without the incorporation of population-level processes. Melanoplus grasshoppers radiated in the sky islands of the Rocky Mountains, and the analysis of divergence between these species suggests that the isolation of populations in multiple glacial refugia was an important factor in promoting speciation. Furthermore, the low estimates of gene flow between the species indicate that reproductive isolation must have evolved rapidly for the incipient species boundaries to be maintained through the subsequent glacial periods and shifts in species distributions.