Abstract: This study aims to estimate the variation characteristics of rainfall erosivity in 10 river basins in China. The daily precipitation data was then selected from 1961 to 2023. Three characteristic values of erosive rainfall were combined, such as the total erosive rainfall, erosive rainy days, and mean erosive rainfall intensity. The trend and cause of annual rainfall erosivity were analyzed by the linear regression and Spearman’s rank partial correlation analysis. The extreme variations were obtained from the trend of annual maximum daily rainfall erosivity and difference of event rainfall erosivity under a 10-year return period during the two periods 1961-2023 and 1961-1990. The recurrence period was statistically fitted for the goodness of fit using the Generalized extreme value distribution (GEV) function and Kolmogorov-Smirnov method. Then, the dangerous situations of rainfall erosivity were comprehensively determined to consider the overall trends and extreme variations in China. At last, the annual rainfall erosivity was predicted using the Hurst index by rescaled range analysis P/S. Some recommendations were given on soil erosion prevention and control in different key erosion regions. The results show that: 1) The ever-increasing trend was found in the annual rainfall erosivity in the Southeast- and Northwest-river basin from 1961 to 2023. But the Southwest-river basin showed a significant decreasing trend. There was no outstanding increasing trend in most other river basins; 2) The positive rate of change was observed in the annual erosive rainfall, rainy days, and mean erosive rainfall intensity in most river basins. Particularly, the significantly increasing trend was found in the mean erosive rainfall intensity in the Yellow, Yangtze, southeast, Pearl, and northwest river basins. The different causes contributed to the trends of annual rainfall erosivity in the 10 river basins. The annual mean intensity of erosive rainfall and/or rainfall was dominated by the trend of annual rainfall erosivity in most river basins. 3) A positive rate of change was found in the maximum daily rainfall erosivity in most parts of China. The number of stations with the extreme maximum historical value was increasing over time, with the peak in the 2010s over that in the 1990s. The ratio of stations with the increasing event rainfall erosivity under a 10-year return period between the two periods 1961-2023 and 1961-1990 to the total was 64.3%; 4) Either the total amount or the extreme of rainfall erosivity was damaged to the soil and water conservation. The rainfall erosivity showed a proportion of 80% of total stations in most parts of the country over the past 60 years. Among them, both the total and extreme rainfall erosivity were the most prominent and widespread in all river basins; 5) The annual rainfall erosivity in most stations in most 10 river basins was predicted to continue the ever-increasing trends during 1961-2023, except for Southwest river basin with ever decreasing trend. The current and future climate conditions are not optimistic for soil erosion control in China. It is very necessary for long-term plans and effective measures, according to the total amount and extreme changes of rainfall erosivity in specific regions. The findings can also provide scientific references for the prevention and control of soil erosion.