Forest residues and logging slash from pre-commercial forest thinning and regeneration harvests are a potential feedstock for bioenergy production but there has been a concern about the impact of residue removal on forest soil C and N. This study aimed to address such by conducting two meta-analyses using the data available from published literature and an independent dataset compiled from the North American Long-Term Soil Productivity (LTSP) study. For the meta-analysis using literature, we categorized forest harvesting and biomass removal into i) no harvest control, ii) bole-only (BO, partial or clearcut) regular harvests, iii) BO with partial removal of logging slash and/or O horizon (BO+Removal), iv) whole tree harvests (WTH), and v) WTH with slash and O horizon removal (WTH+Removal). Accordingly, we compiled soil C and N data and key statistics (e.g., standard deviation) from 142 scientific articles published since 1979. We compared the results from this meta-analysis with data from 22 installations of the LTSP study where three levels of organic matter removal - BO, WTH, and WTH plus forest floor (+FF, O horizon) removal - as well as an additional vegetation control (+VC) were measured for two decades in either completely randomized or randomized block design. In the literature meta-analysis, BO+Removal (-19.2%), WTH (-15.4%) and WTH+Removal (-24.9%) contained significantly less soil C than no-harvest controls across combined soil depths, while BO had no difference. Within individual mineral soil horizons, only BO+Removal and WTH+Removal treatments contained significantly less carbon than controls. There was a high degree of heterogeneity in treatment response between studies in the literature. The analyses from the LTSP dataset showed no significant difference in combined soil depths for WTH or WTH+VC relative to BO harvest, but there was significantly less soil C in BO+VC (-3.6%), WTH+FF (-8.5%) and WTH+FF+VC (-15.3%). These treatment effects declined over time since harvest, particularly the most intensive treatments. Soil N results largely mirrored soil C in both meta-analyses with smaller estimated effects for most treatments at equivalent depths (except for WTH+Removal and WTH+FF+VC, which remain about the same). There were no significant differences in soil N for combined soil depths between WTH and no-harvest control (in the literature analysis) or BO harvest (for both analyses). Since the most severe losses of soil C and N involved FF removal, WTH that accounts for modest removals (<80%) of harvesting residues may provide a sustainable source of biomass for bioenergy production without additional soil impacts compared to BO harvesting practices.