Abstract: Environmental quality standards for surface water of China(GB3838-88)acted as training samples,14 indexes including oxygen-balance parameters,nutrient elements,heavy metal ions and oils were selected,and water quality classification for the mainstream and major tributaries of the Yangtze River was studied with neural network-BP model.The results were combined with GIS digital graphs and indicated that the water quality situation of the mainstream and most tributaries is good,the Tuo River,Great Canal and Taihu Lake are polluted heavily,and the water quality of urban areas of the Yangtze River mainstream are contaminated.