摘要：When the impoundment of the Three Gorges reservoir reaches a water level of 175 m, the reservoir banks along the main river course and its tributaries will have a total length of 5,311 km. Among them, the rock banks are 213 km, and the banks of unconsolidated soil and rock mixture are 249 km. Bank collapse is one of the main engineering geological problems in the Three Gorges reservoir area. In this chapter, using an extensive amount of data on bank collapse obtained through geological investigation, field measurement of the bank collapse parameters, and simulation analysis, we systematically explored the bank collapse types, parameters, and bank collapse prediction methodology. In the Three Gorges reservoir area, the typical bank collapses are wash and abrasion, toe-erosion collapse, rock break-off and slides, and landslide. The evolutionary processes, the distribution features, and the corresponding geological conditions of these bank collapse types are analyzed as well. Eigenvalue angles of the rock and soil masses, including underwater deposit eigenvalue angle θ, wash-abrasion eigenvalue angle α, and steady above-water eigenvalue angle β, are fully discussed in the chapter. Based on the exploration of the correlation between bank collapse prediction parameter and the composition of the rock and soil mass and the correlation between mechanical features and the hydrodynamic conditions in the reservoir, an innovative prediction method that is adaptive to mountain-river reservoirs like Three Gorges reservoir area, Bank-Slope Structure Prediction Method (BSSPM), is proposed. The analogical study of three water storage reservoirs for many years, a comparison of the predicted results with that of the actual happenings, shows that BSSPM has a good adaptation for bank collapse prediction of mountain-river reservoirs.