The aim of this study was to develop valuable model of the interaction between low-power tractors wheel and deformed ground as well as to optimize tractor performance on silty clay loam. The relationships between traction force as well as traction efficiency and soil moisture, soil compaction, horizontal deformation, and vertical load were the subject of investigation. The research was carried out in the laboratory conditions. The two soft computing techniques of mathematical modeling were used: multilayer perceptron and radial basis function neural network. The more efficient model was obtained by multilayer perceptron. For the model with traction force as the output parameter the coefficient of determination was equal to 0,963 (MLP model) and 0,907 (RBF model). For the model with traction efficiency as the output parameter the coefficient of determination was equal to 0,986 and 0,944, respectively. Using the MLP model, the sensitivity analysis was conducted. The highest relative influence on traction force was observed for vertical load, in the case of traction efficiency, horizontal deformation is the most important parameter. For both dependent variables the lowest influence was calculated for soil compaction. The optimization of tractive properties requires generally high horizontal deformation, average soil moisture and high soil compaction. High vertical load is necessary for traction force maximization and relatively low for traction efficiency optimization.