Abstract: Spatial distributions of the monthly–seasonal precipitation in China are similar to the atmospheric circulation teleconnections. Based on the summer precipitation data of China in recent 57 years, this paper investigates the main spatial mode characteristics and interdecadal changes of the spatial teleconnection pattern of the summer precipitation in China. Further, this study evaluates and improves the prediction ability of BCC_CSM, ECMWF_ SYSTEM4, and NCEP_CFSV2 climate models for the summer precipitation in China. Results show that there are four significant spatial teleconnection patterns of summer precipitation in China: (1) North China: lower reaches of the Yangtze River, (2) East China: Central and Northern China, (3) South China: Yangtze River basin, and (4) Southwest China: Central Northeast China. The climate models can only predict the large-scale precipitation distributions; however, the precipitation remote correlation between different regions cannot be well predicted, and there are many false teleconnections. To improve the precipitation prediction ability of the model, a correction scheme is constructed by taking the precipitation teleconnection pattern in the actual situation as the constraint condition to correct the precipitation teleconnection pattern’s distribution in the model. Results show that the correction can effectively improve the model’s ability for predicting the precipitation of Central Northeast China and the lower reaches of the Yangtze River. Four-year return test results show that the average anomaly consistency rate of model prediction increased from 47% to 58%, the root mean square error decreased from 153 to 120 mm, and the average PS (Prediction Score) score increased from 64 to 73.