This work is committed to evaluate spatial inference methods designed for modeling the susceptibility to mass movements in the municipality of São Sebastião (SP) and test the effectiveness of including geomorphometric variables (vertical and horizontal curvatures) in the models. We compared three methods of spatial inference: Boolean, Bayesian, Fuzzy Gama. These methods were tested with five variables (geomorphology, geology, pedology, land use, and slope) and then with seven variables (the former five ones with the addition of vertical and horizontal curvatures). The Boolean method did not allow a detailed classification of susceptibility classes in both cases (with five and with seven variables). The Fuzzy Gama method presented greater flexibility in identifying susceptibility areas and in generating distinct scenarios for both types of models. This was made possible by manipulating the gama index values. The addition of curvatures in the model allowed a better performance and yielded superior results. The Bayesian inference effectively used only slope (in the case of five variables) as an evidence, and slope and horizontal curvature (in the case of seven variables). This method was not satisfactory in discriminating the classes of susceptibility to mass movements.