Abstract: Risk assessment is an important means of disaster prevention and mitigation in mountainous areas in the southwestern China. However, most current evaluation systems use models of shallow level and with single indicator. To address this problem, the authors in this paper proposed the AHP-evidence weight coupling model to conduct risk issessment of geological hazard in Maerkang. ArcGIS hydrological analysis and human-computer interaction were combined to optimize the division of slope units. The study area was divided into 5 695 slope units, and 8 index factors were selected, including rainfall, stratigraphic lithology, slope gradient, slope aspect, distance from water system, distance from fault, and distance from road. AHP-evidence weight coupling model was used to respectively solve the index factor weight value and the contribution of the subordinate intervals to geological hazard events, and the geological hazard risk zoning map of the study area was obtained. The results show that high-risk areas and extremely high-risk areas account for 14.96% and 8.46% of the total area respectively, and these areas are mainly concentrated in residential areas on both sides of the water system. The AUC is 0.78, and the model evaluation accuracy is good. The research results could provide references for the risk assessment of regional geological hazards and have guiding significance for the preliminary work of geological hazard investigation in the southwestern mountainous areas.