There are large uncertainties in predicting soil organic carbon (SOC) in response of future changing climate and human activities. This study estimates SOC stocks under future changing land use and climate conditions in Northeast China using a state-of-the-art digital soil mapping technique. A total of 487 soil samples and 12 environmental variables from 37 landscape units (derived from soil, topography, climate, and human activity data) combined with boosted regression trees (BRT) and random forest (RF) models are first used to map the topsoil (0–20 cm) SOC stocks in Northeast China in 2015. The primary environmental variables influencing the variability of SOC stocks are mean annual temperature, elevation, mean annual precipitation, and land use. We then applied the space-for-time substitution method in conjunction with the BRT model to predict the spatial distribution of SOC stocks under future (the 2050s and 2090s) climate and land use scenarios. SOC stocks under the scenarios of shared socioeconomic pathways (SSP245) and SSP585 (average and upper estimate of the increase in atmospheric greenhouse gases for that time) decreased by 1.5% and 4.5% in the 2050s, respectively, compared with 2015 (5293 Tg C). For the 2090s, the SOC stocks under the SSP245 scenario increased by 1.9%, and those under the SSP585 scenario decreased by 0.4%. The SOC stocks in both future periods are mainly stored in farmlands and forests, accounting for 90% and 92% of the total SOC stocks, respectively. Our high-resolution estimated SOC maps provide a scientific basis for optimizing ecological management in Northeast China.