As soil is a natural resource not always renewable, the risk characterization of contaminated soils is an issue of great interest. Artificial Intelligence (AI), based on Decision Support Systems (DSSs), has been developed for a wide range of applications in contaminated soil management. Decision trees have already shown to be easy to interpret and able to treat large scale applications. Fuzzy logic gives an improvement in the perturbations and the variance of the training data, due to the elasticity of fuzzy set formalism. In this study, we have developed a classificatory tool applied to characterize contaminated soil in function of human and environmental risks. Knowledge engineering for constructing the Soil Risk Characterization Decision Support System (SRC-DSS) involves three stages: knowledge acquisition, conceptual design and system implementation. A total of 26 parameters were divided into three groups to facilitate the configuration of the expert system: source attributes, transfer vector attributes, and local properties. Sixteen case studies were evaluated with the SRC-DSS. In comparison with other techniques, the results of the current study have shown that SRC-DDS is an excellent tool to classify and characterize soils according to the associated risk.