In North American boreal forests, woody biomass is increasingly becoming an attractive bioenergy feedstock as it represents a more renewable, and low carbon alternative to fossil fuel. However, concerns over the ecological sustainability of an intensification of biomass harvesting over the long term remain. We assessed the effects of a gradient of biomass removal established 20 years ago in five jack pine (Pinus banksiana) stands across northeastern Ontario on the taxonomic and functional structure of ground-dwelling Collembola and Oribatida communities. Essential to soil functioning, these two major taxa of mesofauna are dependent on microhabitats and food supply such as fungi provided by woody debris. Three treatments were considered including stem-only harvesting, whole-tree harvesting (stem, tops and branches removed), and blading (whole-tree harvesting plus removal of stump and forest floor). Adjacent uncut mature (±90 year-old) stands were considered as reference state conditions (i.e. the endpoint of stand development following stand replacing disturbance). Soil mesofauna were collected and environmental variables measured across all treatments. We identified species of both taxa and measured a suite of functional response traits such as body length and reproduction strategy. Compared to mature uncut forests, soil mesofaunal communities remained modified 20 years after biomass harvesting, notably in the most intense practice (blading). Treatment effects were more evident in Oribatida communities both taxonomically (lower density, biomass, species diversity and shifted composition) and functionally (lower diversity and modified trait composition resulting from fewer surface-dwelling, fast-dispersing and micro-detritivorous species) than in Collembola communities. This taxa-specific response may reflect generally shorter lifespans, higher reproduction rates and faster dispersal of Collembola than Oribatida. Incomplete recovery of mesofauna was consistent with persistent modifications of soil environmental conditions in harvested plots, notably after blading. Modifications included a reduced organic cover (ground vegetation, mosses and woody debris) as well as lower organic soil thickness and moisture, which likely resulted in fewer suitable microhabitats for many species. Using complementary taxonomic and trait-based approaches to highlight the underlying mechanisms of mesofaunal responses, our study revealed that recovery is incomplete within 20 years after intensive biomass removal in these boreal conifer-dominated stands, and is likely linked to stand development and associated processes. As a result, longer-term monitoring will be required to track mesofaunal community recovery through these later developmental stages.