Collapse mound, a serious soil erosion phenomenon in granite areas of South China, occurs mainly on mound or hill slopes under the interaction of water and gravity, causing severe destructions and threats in a wide range, including Hubei, Hunan, Jiangxi, Anhui, Fujian, Guangdong, and Guangxi, mainly south to the Yangtze River, and hence a grave impact on the economy in these hilly regions. So far a lot of research work has been done on mechanisms of how mount collapses occur. In the aspect of water regime in collapse mounds, related scientists have analyzed permeability of the colluvial deposits from collapse mounds, revealing that colluvial soil reaches the level of steady permeation in a relatively short time, and also studied soil permeability of collapse mounds at different depths. Analysis from the angle of initial infiltration rate, stable infiltration rate, average infiltration rate, and infiltration angle shows that permeability declines steadily from the red earth layer down to the detritus layer. Therefore, this article proceeds from collapsing of mounds and water regime of the soil per se, with a view to elaborating water movement processes during the collapse of mounds or cliffs, exploring soil water characteristic curves of collapse mound profiles, including topsoil layer, red earth layer, streaked layer, and detritus layer, and their fitting processes with equations, and analyzing the mechanisms of how collapse mounds occur from the angle of soil moisture and pority. The study adopted field sampling and in-lab analysis together. Two typical profiles of collapse mounds in Southeast Hubei, Wuli and Yanglong were selected as subjects of the study. Soil water characteristic curves in various soil layers were plotted with the aid of a CR21G high-speed thermostat centrifuge of the Hitachi Corp. and on the basis of the soil water characteristic curves, soil porocities of the various soil layers were analyzed. Meanwhile, equations were screened for fitting the various soil water characteristic curves of the profiles. Results showed as follows: (1)In the Wuli and Yanglong profiles, soil release rate varied regularily with soil water suction rate. In the streaked soil layer and sandy soil layer, soil water release rate was high when their soil water suction was low, and the soil water characteristic curves in various soisl layers all leveled off. The topsoil and the red earth layers were higher than the streaked and detritus soil layers. (2)Based on the field-measured values of the water characteristic curve and the calculated equivalent porecity, pore size distribution in the profiles of collapse mounds was studied. It was found that the two profiles followed a similar law in soil pore size distribution, that is, the proportion of large pores increased while that of capillary pores decreased with soil depth going down from the topsoil layer to the detritus layer. The increase in number of large pores created a favorable condition for water movement, thus enhancing the probability of collapsing soil erosion. (3) The van Genuchten equation and the Gardner equation were tested to fit the curves, and the fitting was evaluated. It was found that the van Genuchten equation was better to fit the measured data of the soil moisture characteristic curve of the topsoil and red earth layers with relatively little deviation, while the Gardner equation was better to fit those of the steaked soil layer and the debris layer. On the whole, the van Genuchten equation is higher in fitting accuracy. Evaluation using the residual square sum method indicates that generally speaking, it is advisable to use the van Genuchten equation to fit soil water characteristic curve of granite soil.