Soil aggregate stability is a key parameter influencing soil fertility, environmental problems, and soil sensitivity to erosion. It is well established that aggregate stability changes over time. However, there is a lack of information considering the dynamics of aggregate at a shortterm step in the field. In addition, aggregate stability is usually determined by using soil samples collected from the plough layer, while soil erosion occurs in the surface soil. The difference between the soil crust and plough layer is usually ignored. The objectives were to investigate the variations in aggregate stability under rainfall event, identify the factors controlling these variations of stability, and compare the difference of aggregate stability between soil crust and subsurface soil. A field experiment was conducted on a cropland with typical yellow soil in Three Gorges Reservoir area. Sampling was performed after rainfall events. Aggregate stability was measured for both crust soil and subsurface soil by the Le Bissonnais (LB) method. Rainfall amount, rainfall intensity, air temperature, soil temperature and soil water content at the time of sampling were measured as potential explanatory factors. The results showed that aggregate stability varied significantly (P<0.05) under rainfall event for both the soil crust and subsurface soil. During the experiment, the mean weight diameter (MWD) obtained under fast wetting, slow wetting, and prewetting shaking treatments was varied between 0.73mm and 2.13mm, 1.70mm and 2.90mm, and 2.27mm and 3.16mm, respectively, for the soil crust. While MWD of subsurface soil was varied between 0.70mm and 1.27mm, 1.69mm and 2.83mm, and 2.53mm and 2.95mm, respectively. Great differences between the aggregate stability of soil crust and subsurface soil were found. And the aggregate stability of soil crust was significantly higher than that of surface soil. For the soil crust, accumulative rainfall amount, rainfall intensity, air temperature, soil temperature and soil water content at the time of sampling greatly controlled the variations of aggregate stability in fast wetting and slow wetting treatments. The large differences in aggregate stability between soil crust and the plough layer implied large differences in soil erodibility. And the soil erodibility was varied greatly in the field over several rainfall events.