Subspecies designations within temperate species' ranges often reflect populations that were isolated by past continental glaciation, and glacial vicariance is believed to be a primary mechanism behind the diversification of several subspecies of North American cervids. We used genetics and the fossil record to study the phylogeography of three moose subspecies (Alces alces andersoni, A. a. gigas, and A. a. shirasi) in western North America. We sequenced the complete mitochondrial genome (16,341 base pairs; n = 60 moose) and genotyped 13 nuclear microsatellites (n = 253) to evaluate genetic variation among moose samples. We also reviewed the fossil record for detections of all North American cervids to comparatively assess the evidence for the existence of a southern refugial population of moose corresponding to A. a. shirasi during the last glacial maximum of the Pleistocene. Analysis of mtDNA molecular variance did not support distinct clades of moose corresponding to currently recognized subspecies, and mitogenomic haplotype phylogenies did not consistently distinguish individuals according to subspecies groupings. Analysis of population structure using microsatellite loci showed support for two to five clusters of moose, including the consistent distinction of a southern group of moose within the range of A. a. shirasi. We hypothesize that these microsatellite results reflect recent, not deep, divergence and may be confounded by a significant effect of geographic distance on gene flow across the region. Review of the fossil record showed no evidence of moose south of the Wisconsin ice age glaciers ≥ 15,000 years ago. We encourage the integration of our results with complementary analyses of phenotype data, such as morphometrics, originally used to delineate moose subspecies, for further evaluation of subspecies designations for North American moose.