Incomplete lineage sorting can genetically link populations long after they have diverged, and will exert a more powerful influence on larger populations. The effects of this stochastic process can easily be confounded with those of gene flow, potentially leading to inaccurate estimates of dispersal capabilities or erroneous designation of evolutionarily significant units (ESUs). We have used phylogenetic, population genetic, and coalescent methods to examine genetic structuring in large populations of a widely dispersing bat species and to test hypotheses concerning the influences of coalescent stochasticity vs. gene flow. The Mexican free-tailed bat, Tadarida brasiliensis mexicana, exhibits variation in both migra- tory tendency and route over its range. Observations of the species’ migratory behaviour have led to the description of behaviourally and geographically defined migratory groups, with the prediction that these groups compose structured gene pools. Here, we used mtDNA sequence analyses coupled with existing information from allozyme, banding, and natural history studies to evaluate hypotheses regarding the relationship between migration and genetic structure. Analyses of molecular variance revealed no significant genetic structuring of behaviourally distinct migratory groups. Demographic analyses were consistent with population growth, although the timing of population expansion events differs between migratory and nonmigratory populations. Hypotheses concerning the role of gene flow vs. incomplete lineage sorting on these data are explored using coalescent simulations. Our study demonstrates the importance of accounting for coalescent stochasticity in formulating phylogeographical hypotheses, and indicates that analyses that do not take such processes into account can lead to false conclusions regarding a species’ phylogeographical history.