Abstract: Debris flow has great destructive power, especially in mountainous areas, which seriously threatens the safety of human life and property. Therefore, it is of great significance to study the susceptibility assessment of debris flow in mountainous areas for spatial planning of national land, disaster prevention and mitigation, and formulating reasonable measures to prevent and control debris flow. Taking the region of Dongchuan where debris-flows frequently occur as an example, this paper adopts two machine learning algorithms of random forest (RF) and support vector machine (SVM), and takes the watershed unit as the assessment unit. Based on the identification of debris flow points, 15 index factors are selected to build the assessment model of debris flow susceptibility in mountainous areas, the weight of each inducing factor is evaluated, and the prediction effect of the model is compared and analyzed. The results show that the two machine learning algorithms combined with watershed units have good accuracy and stability, but the ACC and the AUC values of random forest (RF) model are 0.8333 and 0.9299, which are better than those of support vector machine (SVM) model, namely 0.7222 and 0.8515. The random forest (RF) model is more suitable for the assessment of debris flow susceptibility in mountainous regions.