Applying a data mining tool used regularly in digital soil mapping, this research focuses on the optimal inclusion of predictors for soil–landscape modeling by utilizing as wide of a pool of variables as possible. Predictor variables for digital soil mapping are often chosen on the basis of data availability and the researcher's expert knowledge. Predictor variables commonly overlooked include alternative analysis scales for land-surface derivatives and additional remote sensing products. For this study, a pool of 412 potential predictors was assembled, which included qualitative location classes, elevation, land-surface derivatives (with a wide range of analysis scales), hydrologic indicators, as well as proximal and remote sensing (from multiple sources with a variety of resolutions). Subsets of the full pool were also examined for comparison. The performance for the models built from the different starting predictor pools was analyzed for seven target variables. Results suggest that models with limited predictor pools can substitute other predictors to compensate for the missing variables. However, a better performing model was always found by considering predictor variables at multiple scales. Compared with baseline subsets with the most commonly used predictors for digital soil mapping at a single scale, the use of multi-scale predictor variables produced an improvement in model performance ranging from negligible to a 70% increase in the adjusted R2. Although the scale effect of the modifiable area unit problem is generally well known, this study suggests digital soil mapping efforts would be enhanced by the greater consideration of predictor variables at multiple analysis scales.