Development of adequate GIS-based soil erosion risk models provides scientific justification for water-erosion arrangement used in agricultural landscapes. The estimated soil loss is highly influenced by model input data. The aim of the paper is the impact assessment of various digital terrain models (DTMs) on soil erosion risk modeling. The elevation datasets used in the study include five DTMs from various sources, viz., ALOS World 3D, 30-m and 90-m SRTM DTMs, DTM derived from Sentinel-1 data with SNAP, and 10-m DEM based on 1:10,000 scale topographic maps. The DTMs in soil erosion modeling are utilized for the estimation of the LS factor constituting slope steepness factor (S-factor) and slope length factor (L-factor); therefore, their role is crucial. For a detailed analysis of the above-mentioned models, a study area with quite contrasting relief was selected. It is located within the Minsk uplands; the relief is hilly with prevailing elevations of 230–270 m. The research methodology comprised DTM generation from Sentinel-1 using InSAR techniques available in SNAP, comparative elevation and slope value assessments, topographical factors evaluation of the RUSLE equation: slope length (L) and slope steepness (S) and a detailed analysis of the total LS factor values. A statistical analysis of the elevation datasets indicates a significant similarity in values, as evidenced by the high Pearson correlation coefficients calculated for all DTM pairs (> 0.84). However, when comparing the slope values computed by different terrain models significant variability in their distribution is observed. The Pearson coefficients calculated for all pairs of slope values decrease compared to those for the terrain elevations (DTM values) and range from 0.08 to 0.47. Both the values of the LS factors and their spatial patterns vary depending on the basic digital terrain model used. The values of the LS factors do not change uniformly with increasing DTM accuracy and detail. When compared to the most precise 10-m topo DEM at a scale of 1:10,000, the relative error of the mean LS factors was less than 15% when calculated from the 30-m SRTM DTM, and approximately 60% for ALOS DTM.