Soil-landscape models have prediction errors that can be reduced by using auxiliary soil data. However, standard soil surveys using auger hole and laboratory analysis encounter both methodological and economical constraints because of, for example, the short-range variability of soils and the expensive field work. In the present study, the objective was to test the use of auxiliary geophysical data to improve a soil-landscape model for two Digital Elevation Model (DEM) resolutions (10 and 50 m). The study site was an agricultural parcel comprising an entire hillslope of a small (0.6 km2) catchment in Normandy, France. Prediction models based on multiple regression and co-kriging techniques were established using topographic, soil and geophysical data. For high DEM resolution (10 m), soil-landscape models based only on surface features seemed to be efficient. The use of additional geophysical or soil data improved the prediction quality slightly. For coarser DEM (50-m resolution), the prediction quality of models established using only terrain attributes was faulty. The soil hydromorphic prediction could be improved greatly by the use of auxiliary geophysical data. In this case, the improved accuracy was similar to that obtained by high soil density data (50 observations/ha). Finally, we discuss the use of both geophysical and topographical data in order to describe better the spatial distribution of soils.