Abstract: Taking Yanzi River Basin as an example, this paper studied the effect of influencing factor classification results on landslide susceptibility mapping. According to the historical landslides data and geological environment characteristics, this paper used elevation, slope, aspect, plane curvature, profile curvature, stratigraphic lithology, distance to faults, distance to roads, normalized difference vegetation index, topographic wetness index and standardized precipitation index as the influencing factors. With different classification data, the classification results are different, which will produce different susceptibility maps. Therefore, for continuous factor classifica￣tion, this paper used the landslide points and all points in the study area as the classification data, respectively. The optimal classification number of continuous impact factors was determined by the inflection point method, and the corresponding classification interval was determined by Fisher-Jenks algorithm. Discrete impact factors were classified according to the actual situation. After influencing factor classification, support vector machine method was used to construct susceptibility mapping models, and the receiver operating characteristic(ROC)curve was used to evaluate the model performance. The results show the inflection point method and Fisher-Jenks algorithm can be well applied to the classification of continuous influencing factors. Compared with using all points in the entire study area to determine the factor classification criteria, the classification effect of using landslide points is better and the susceptibility map performs better. The areas under ROC curve corresponding to the two methods are 0.82(using all points in the study area as classification data) and 0.87(using all landslides points as classification data),respectively. The susceptibility maps of the Yanzi River Basin drawn in this paper can be used as a reference for disaster prevention and reduction.