The purpose of the present study is to model the hydrocarbon resources potential mapping using geographic information systems (GIS). The presented method is based on petroleum system concepts; therefore, petroleum system elements were used to define criteria for petroleum potential mapping. Several statistical methods can be used to effectively predict potential areas for hydrocarbon resources. In this study, two statistical methods were used (frequency ratio and evidential belief functions) to predict the potential distribution of petroleum resources in the study area. A case study in the Red River – Red River petroleum system of the Canadian Williston Basin in southeastern Saskatchewan in Canada is proposed to assess the feasibility of this new modelling technique. The accuracy of the hydrocarbon potential maps was evaluated by success rate and prediction rate efficiency curves. The resultant petroleum potential maps resulted in delineation of high-potential zones occupying about 15% of the study area. The validation results showed that the prediction rate for the best model is 88.14%. This study was carried out at a regional scale; therefore, the results can be used to guide exploration works at early stages.