Abstract In many regions of the world, forest management has shifted from practices emphasizing timber production to more sustainable harvesting that integrates ecological values, including maintenance of biodiversity, wildlife habitat, and ecological goods and services. To this end, management strategies emphasize retention of stand structures that meet the needs of forest-obligate wildlife species and enhance connectivity across landscapes. However, little is known about the effects on arboreal rodents of varying the amount or spatial distribution of retained structures. We quantified the responses of northern flying squirrels ( Glaucomys sabrinus ) to retention harvests of varying levels (15%, 40%, 75%, and 100% of original basal area) and spatial patterns (trees uniformly dispersed vs. aggregated in 1-ha patches), using six experimental treatments replicated at three locations in southwestern Oregon and Washington. Relative abundance of northern flying squirrels decreased following harvest; minimum number of squirrels known alive (MNKA) in the control (100%) and 75% retention treatment was significantly higher than in the 15% or 40% treatments. In mixed-effects regression models, MNKA increased with treatment-unit basal area and amount of surrounding mature (>80-year-old) forest, suggesting that squirrel abundance was influenced by local structure and landscape-scale variables. However, only basal area contributed to best-fit models of reproductive female abundance. Our results suggest a threshold response of northern flying squirrels to green-tree retention somewhere between 40% and 75% that is likely to be influenced by the spatial pattern of retention and landscape context. This study underscores previous conclusions that northern flying squirrels are sensitive to logging at both local and larger landscape scales, and demonstrates the current minimum retention standard of 15% will not provide suitable habitat for this species.