The distribution of summerhalf year rainfall in the Pearl River basin is obtained with the monthly rainfall data of April to September from 1954 to 2003 in the basin, the antecedent Pacific SST (Sea Surface Temperature) field and 500 hPa height field, by means of the principalcomponent〖CD*2〗stepwiseregression prediction model. The results show that the first 25 principal components of summer half year rainfall can well describe the variation of summer half year rainfall in the basin. The distribution prediction of summerhalf year rainfall from 2004 to 2006 in the basin shows that the prediction results are similar to the real distribution. The prediction factors taking from the principal components of the 500 hPa height field and the SST field over the North Pacific contain a large amount of useful information for rainfall prediction. The method has certain significance for operational prediction.