Comprehensive forest carbon accounting requires reliable estimation of soil organic carbon (SOC) stocks. Despite being an important source of carbon, limited information is available on SOC stocks in global forests, particularly in forests in montane regions such as the central Himalayas. The availability of rigorously measured new field data allowed us to accurately estimate forest soil organic carbon (SOC) stocks in Nepal, addressing a previously existing knowledge gap. Our method involved modeling forest SOC estimates using covariates related to climate, soil, and topographic location. Our quantile random forest model resulted in high spatial resolution prediction of Nepal's national forest SOC stock along with prediction uncertainties. Our spatially explicit forest SOC map showed high levels of SOC in high-elevation forests and a significant underrepresentation of these stocks in global-scale assessments. Our results offer an improved baseline of the distribution of total carbon in central Himalayan forests. Reference maps of predicted forest SOC and associated errors, together with our estimate of 494 million tons (SE = 16) of total SOC in the topsoil (0–30 cm) of forested areas in Nepal have important implications for understanding the spatial variability of forest SOC in mountainous areas with complex terrain . Accurate and reliable estimation of national soil organic carbon (SOC) stocks is critical for reporting in the context of the UN's Reducing Emissions from Deforestation and Forest Degradation (REDD+) 1 and several other initiatives. Assessing carbon stocks in forest ecosystems requires an understanding of tree biomass, belowground biomass and soil carbon, as well as fluxes. (Shiva Khanal, Rachael H. Nolan, Belinda E. Medlynová, Matthias M. Boer, Scientific Reports)