Retention forestry is a common practice for biodiversity conservation in forests managed for wood production. Retention forestry often leaves unharvested patches of trees that vary in size and spatial pattern but experiments evaluating the effects of different retention patch configurations at a constant level of retention are lacking for many regions and taxonomic groups. We implemented an experimental study in clearcut conifer stands with retention across the U.S. Pacific Northwest region. The study consisted of five stand-level (11-55 ha) experimental treatments each replicated 10 times within a randomized complete block design, resulting in 50 treated stands. Retained tree density was comparable across treatments but size, number, and location (upland or riparian) of patches within stands varied among the five treatments. Within experimental treatments, we measured small mammal (<1kg) species and functional trait (i.e., body size, diet, activity stratum) richness in retention patches, surrounding harvested portions of stands, and nearby unharvested stands. We evaluated species and functional trait richness by treatment using generalized linear mixed-effects models and species-specific responses to retention placement using a community occupancy model. We obtained repeat captures of 21 species of small mammals but found limited evidence of a treatment effect on species richness, and no differences in functional trait richness. Species richness was highest where all retained trees were aggregated into one patch placed adjacent to a forested riparian buffer (mean = 6.6 species, 95% CI = 5.7-7.5), and lowest in the treatment containing one retention patch in the upland portion of a harvested stand (mean = 4.7 species, 95% CI = 3.8-5.6). Furthermore, estimates of species richness within retention patches of harvested stands (i.e., not considering species in harvested areas) did not differ among treatments, indicating that the slightly elevated species richness in riparian-associated retention results from 1-2 species in these patches that do not occur in adjacent harvested portions of each treated stand. Patch occupancy of several species was higher in riparian patches than harvested portions of the treated stands, and fewer species had higher occupancy in upland patches compared to harvested portions of treated stands. Our results indicated that at retention densities currently required in Oregon and Washington, the location of retention patches had a small influence on stand-scale measures of small mammal diversity, but local increases in species richness may be obtained by retaining trees adjacent to riparian buffers.