AbstractThe spatial distribution of soil organic carbon is an important factor in land management decision making, climate change mitigation and landscape planning. In Scotland, where approximately one‐quarter of the soils are peat, this information has usually been obtained using field survey and mapping, with digital soil mapping only carried out recently. Here a method is presented that integrates legacy survey data, recent monitoring work for peatland restoration surveys, spatial covariates such as topography and climate, and remote sensing data. The aim of this work was to provide estimates of the depth, bulk density and carbon concentration of Scotland's soils in order to allow more effective carbon stock mapping. A neural network model was used to integrate the existing data, and this was then used to generate a map of soil property estimates for carbon stock mapping at 100‐m resolution over Scotland. Accuracy assessment indicated that the depth mapping to the bottom of the organic layer was achieved with an r2 of .67, whereas carbon proportion and bulk density were estimated with an r2 of .63 and .79, respectively. Modelling of these three properties allowed estimation of soil carbon in mineral and organic soils in Scotland to a depth of 1 m (3,498 megatons) and overall (3,688 megatons).Highlights Scotland's soil organic carbon was mapped using a digital soil mapping approach. This provides a high‐resolution map available for scientists, regulatory bodies and policymakers. The method largely agreed with previous work but improved the spatial resolution of the mapping. Significant soil carbon stocks are held in both organic (peat) and non‐peat soils.