We evaluated the two-year effects of variable-retention harvest on chipmunk (Tamias spp.) abundance () and habitat in mature coniferous forests in western Oregon and Washington because wildlife responses to density/pattern of retained trees remain largely unknown. In a randomized complete-block design, six treatments were applied to 13-ha units at three sites (blocks): four retention levels of original basal area (BA) in an aggregated tree pattern (100, 75, 40, and 15%) and two retention levels in a dispersed tree pattern (15 and 40%). Log-yarding method differed at each site (suspension cable, shovel-loader, or helicopter). We used an information-theoretic approach to compare six candidate regression models for their ability to predict treatment responses of chipmunk and associated habitat variables. Chipmunk had a positive linear relationship with retention level that predicted a 50% reduction in abundance as % BA retention decreased from 100 to 15% (R² = 0.36). Disturbed soil cover was strongly related to the interaction of retention level and block (i.e., yarding method and other site-level differences) (R² = 0.82), and the model predicted disproportionately greater disturbed area for cable yarding (16%) than for shovel (10%) or helicopter (6%) methods as retention decreased from 100 to 15%. Chipmunk had a negative linear relationship with disturbed soil cover that predicted a 70% reduction in the species' abundance as disturbed area increased from 0 to 16% (R² = 0.53). Retention level and yarding method are important considerations when planning harvesting operations because of their potential impacts to small mammal populations.