Post-harvest recovery of biodiversity is one of the important goals in modern forestry. A variable retention (VR) approach has been of particular interest in North America because it promotes rapid faunal recovery while minimizing the negative lasting impacts of logging on the natural fauna. We studied responses of rove beetles (Coleoptera: Staphylinidae) to a broad range of retention harvests (2, 10, 20, 50, and 75 % retention) in comparison to uncut controls as part of the Ecosystem Management Emulating Natural Disturbance (EMEND) experiment in the boreal mixedwood forest of western Canada. We sampled beetles using pitfall traps 1, 2, 11, and 16 years post-harvest in replicated (n=3) stands representing four cover types (deciduous-dominated, deciduous with spruce understory, mixed, and coniferous-dominated). We collected 74,263 individuals distributed across 99 species (excluding Aleocharinae). Estimated species richness was highest in clear-cuts until year 11, but by year 16 species richness was similar among treatments. Species composition initially varied strongly in relation to the intensity of harvest treatments, but overall variation decreased with time, and by year 16, species composition overlapped among most treatment combinations. Assemblages recovered more quickly in early successional (deciduous-dominated) than in late successional (mixed and conifer-dominated) stands. Overall, our results show that rove beetle assemblages in stands harvested to all VR prescriptions converged more rapidly toward those in fire-origin mature stands than did assemblages in clear-cuts over the first 16 years post-harvest. Thus, it demonstrates that even modest levels of forest retention can facilitate the recovery of staphylinid assemblages in managed landscapes.