Based on the minute-per-minute rainfall observation data of 32 national meteorological stations in Chongqing City from 2009 to 2018, the characteristics of amount, duration, intensity and erosivity of erosive rainfalls were analyzed to explore the spatiotemporal distribution characteristics and different patterns of erosive rainfalls in the study area. The results show that the amount, duration, intensity and occurrence frequency of the average annual erosive rainfall accounted for 75%, 44%, 55% and 18% of those of the average annual rainfall, respectively. Erosive rainfalls were mainly concentrated in May to September, of which the amount and duration showed a bimodal annual variation; whereas the intensity, occurrence frequency and rainfall erosivity presented a unimodal annual variation. The annual rainfall erosivity ranged from -22% to 27%, with a unimodal distribution. The various characteristic indices of erosive rainfalls showed obvious regional differences at different time scales. Then, principal component cluster analysis (PCA) was used to divide the study area into three types: light, moderate and severe rainfall erosion areas, and the rainfall patterns in the study area were divided into three categories based on K-means clustering, in which Category I was the dominant rainfall type in the region, Category Ⅲ was the rainfall type with the most severe erosivity, and the rainfall amount, intensity, duration and erosivity of Category II were between those of Category I and Category III.