Soil organic carbon (SOC) is the target attribute with the largest number of studies in digital soil mapping in the frame of global initiatives such as GlobalSoilMap that aim to provide maps of key soil properties at national scales. Previous studies have considered terrain-derived variables and other soil properties such as clay to map SOC. Nowadays with the free availability of satellite imagery the mapping of SOC using remote sensing data is being more widely envisioned. However, the capability to predict SOC via satellite imagery in conjunction with soil and terrain data is still unknown in many agricultural regions. In order to map top SOC content over croplands, this study was carried out in the framework of the STEROPES project in two regions in France, Beauce (4838 km2) and Pyrenean piedmont (22177 km2), addressing: (i) for both areas, the use of Sentinel-2 single date images and/or soil moisture maps derived from Sentinel 1 and 2 data (ii) for Beauce only, the constructing of temporal mosaics of bare soil images, the inclusion of Gamma-ray images and terrain-derived covariates in prediction models and the map uncertainties. In both study areas, the prediction performances were influenced according to the date of image acquisition, surface soil conditions (e.g., soil moisture and soil roughness) and the historical context. In Pyrenean piedmont based on a purely spectral method (i.e. using only spectral data from S2 images) using single date images it was possible to predict high SOC contents in specific soil types. In Beauce considering a mixed method (i.e. using S2 spectral data and terrain-derived variables as covariates in the prediction models) using S2 temporal mosaics of bare soil (S2Bsoil) it was possible to determine the most relevant variables in that area and the best periods to elaborate S2Bsoil for predicting SOC.