Abstract: The Yigong Zangpo Basin is characterized by steep terrain and fragmented rock masses, resulting in frequent large-scale collapses and landslides that pose serious risks to hydropower development in the area. This study utilized remote sensing image interpretation and field surveys to identify the spatial distribution patterns of landslides and analyze their controlling factors. Landslide susceptibility was assessed and predicted using the Frequency Ratio(FR)model, the Normalized Frequency Ratio(NFR)model, and a coupled NFR-Logistic Regression(NFR-LR)model. The results indicate that: (1)A total of 705 collapses and landslides were identified, forming high-density clusters in areas such as the Yigong Lake-Xiaqu hydropower station section and the Zhongyu-Jinqiao-Longyan hydropower station section. (2)Favorable conditions for collapses and landslides in the basin include: proximity to rivers within 1 km, elevations below 4500 m, slope gradients steeper than 30°, southeast-, south-, and southwest-facing aspects, topographic relief greater than 450 m, distances to faults between 0.1 km and 4 km, NDVI values above 0.45, road density exceeding 0.1 km·km^-2, and the presence of weaker rock groups. (3)The prediction accuracy(AUC)of all three models(FR, NFR, and NFR-LR)exceeded 0.85. The moderate-to-high susceptibility zones delineated by these models contained 90.92%, 98.30%, and 89.64% of the documented collapses and landslides, respectively. (4)The coupled NFR-LR model achieved the highest prediction accuracy(AUC=0.870). It improved the precision of susceptibility classification, enabling accurate identification of high-susceptibility areas while effectively excluding low-risk zones, thus supporting more efficient disaster prevention decision-making.