Based on the daily precipitation data in four provinces or cities in Southwest China from 1980 to 2017, the spatial anomaly characteristics of precipitation in Southwest China during the rainy season were analyzed using the methods of principal component analysis and rotated principal component analysis. By the regional division, the characteristics of temporal evolution of precipitation in different regions were discussed. Finally, the prediction model was established using the mean generating function, and the prediction experiment was carried out. The results show that the precipitation in Southwest China during the rainy season has a large spatial difference and can be divided into six regions. There are significant inter-annual and inter-decadal characteristics in each region.The precipitation in most regions in southwest is reduced in recent years, and the year with abnormally less precipitation appears more frequently, and the prediction model will achieve good results for both the fitting and prediction of precipitation in the rainy season.