Abstract: The accumulation layer landslides in Qinba Mountain area are abundant, widely distributed and frequently, and the harm caused by them is very serious. Moreover, it is characterized by complex and diverse disaster pregnancy conditions and difficult to obtain some disaster evaluation data. Xiaoling Town, Qinba Mountain, was selected as the research area. The geological hazard field survey was taken as the basis. Combined with the regional characteristics of accumulation landslide, two element types, grid element and slope element, are adopted. The landslide hazard factors were selected according to local conditions, and their correlation was analyzed. Eight factors, including slope, slope height, slope morphology, slope structure type, accumulation layer thickness, distance from road, mining area and fault, are selected as the characteristic factors of accumulation layer landslide. The random forest model method was used to evaluate the landslide susceptibility of the town area. In addition, the accuracy and accuracy of grid element and slope element were verified by frequency ratio, ROC curve, mean value and standard deviation of susceptibility probability of evaluation results. The results show that both evaluation elements have good performance in the re-prediction results, but the overall prediction performance of slope element as evaluation element is higher than that of grid element. In the specific spatial deployment of disaster prevention and control, more detailed reference comes from grid element. The research results have certain theoretical and practical significance for the risk assessment of geological hazards in towns in Qinba Mountains.