Flood and drought caused by summer precipitation anomaly are the most serious disasters in Yellow River Basin, so it is very important to predict the summer precipitation in the Yellow River Basin. Based on summer precipitation observation in the River Basin, and the BCCCGCM model data, a pattern projection downscaling method is used to predict summer precipitation there. The results show that: (1) The prediction result of pattern projection downscaling method is better and more stable than model output prediction for Yellow River Basin. The monthly prediction is better than seasonal prediction. (2) The pattern projection downscaling method produces different predictions for different areas in the Yellow River Basin. In summer (JJA), the better prediction regions are dispersive. But the better prediction regions are collective in monthly prediction. The prediction capability of the method varies with month. (3) The results of independent sample test on the data from 2009 to 2013 show that the prediction of pattern projection downscaling method for Yellow River Basin is better than model output. Especially in June and July, the prediction of downscaling is greatly superior to the model output.