We studied habitat relations of the Prince of Wales flying squirrel (Glaucomys sabrinus griseifrons), an endemic of the temperate, coniferous rainforest of southeastern Alaska, because of concerns over population viability from extensive clear-cut logging in the region. We used stepwise logistic regression to examine relationships between microhabitat use (i.e., captures among traps spaced at 40-m intervals) and 26 vegetative and structural habitat features measured in plots centered on trap stations. Seasonal (spring, autumn) models were created for two old-growth forest types: upland, western hemlock (Tsuga heterophylla)–Sitka spruce (Picea sitchensis) forests (upland-OG); and peatland-scrub–mixed-conifer forests (peatland-MC). Density of trees with diameter at breast height (dbh) >74 cm and abundance of Vaccinium were positively correlated with microhabitat use in peatland-MC during both seasons. During spring and autumn, the odds of capturing a flying squirrel increased by factors of 2.7 and 16.9, respectively, with an increase in mean density of 10 large trees/ha. Microhabitat use of upland-OG during autumn was positively correlated with density of snags with a dbh of 50–74 cm and negatively correlated with percentage cover of understory herbaceous vegetation; microhabitat use during spring was inversely correlated with percentage surface cover of water. At the macrohabitat (13-ha replicate of forest type) scale, large (>74-cm dbh) trees explained 65% of the variation in density between forest types; percent cover of moss and volume of down wood in decay classes I–IV explained 70% and 63–77% of the variation, respectively. Our findings corroborate general patterns reported for western coniferous forests, but suggest that G. sabrinus in temperate rainforests of southeastern Alaska differ ecologically from populations in the Pacific Northwest in important ways.