Considering the uncertainty of indexes for evaluating water quality and the standard of classification,Fuzzy Probabilistic Neural Network Model(FPNN)was proposed by combining the relative membership grade in fuzzy mathematics and Probabilistic Neural Network(PNN).The process of this model was clarified,and the method of establishing the studied data based on relative membership grade matrix was brought forward.Finally the model was applied to the actual water quality evaluation.The result indicates that the proposed method is easy to operate and the outcomes is objective and credible compared with those by integrated evaluating method,attribute recognition model and BP network.