Spatial population structure has important ecological and evolutionary consequences. Little is known about the population structure of snowshoe hares (Lepus americanus), despite their ecological importance in North American boreal forests. We used seven variable microsatellite DNA loci to determine the spatial genetic structure of snowshoe hares near Kluane Lake, Yukon during a cyclic population peak. We sampled 317 hares at 12 sites separated by distances ranging from 3 to 140 km, and used 46 additional samples from Alaska and Montana. The level of genetic variation was high (13.4 alleles/locus, 0.67 expected heterozygosity) and the distribution of alleles and genotypes was not homogeneous across the sites. The degree of differentiation was low among Yukon sites (FST = 0.015) and between Yukon and Alaska (FST = 0.012), but the Montana site was highly differentiated (FST = 0.20). A weak pattern of isolation by distance was found over the Yukon study area, with an indication that local genetic drift may be important in shaping the regional genetic structure. Landscape barriers expected to influence gene flow did not consistently affect genetic structure, although there was evidence for a partial barrier effect of Kluane Lake. The high level of inferred gene flow confirms that snowshoe hare dispersal is widespread, successful and equal between the sexes. A stepping-stone model of gene flow, potentially influenced by the synchronous density cycle, appears to best explain the observed genetic structure. Our results suggest that despite their dramatic fluctuations in density, snowshoe hares in the northern boreal forest have a large evolutionary effective population size and are not strongly subdivided by either physical or social barriers to gene flow.