In light of the problem that how to select the wavelet function facing the prediction of hydrological series by the wavelet based gray model, this paper analyzed the impacts on the accuracy of different wavelet based gray models by taking the measured rainfall of the Anqing Station as an example and using the predicted values of the traditional gray model as a reference. It also took the seven kinds of wavelet functions including Haar, Db, Smy, Coif, Bior, Rbio and Dmey into consideration and made the Root Mean Squared Error (RMSE) and the Nash–Sutcliffe Efficiency (NSE) as the evaluation?indexes to the models. The result shows that different wavelet functions have both positive and negative effects on the accuracy of models, among which db1 wavelet function is the most effective one to improve the accuracy of the model based on Anqing’srainfall series..