Conventional water quality prediction methods have the disadvantages of poor prediction accuracy and high requirements for monitored data, in order to have a better grasp of the future water environment quality of the Xiaohe River Basin over a certain period of time, a BP neural network model was established to predict the concentration of water quality indices. The indices of CODCr and CODMnof six water quality monitoring sections in the basin from January 2017 to May 2020 were selected as the training set and the water quality data from June to August 2020 as the validation set for the model. The simulation results show that the average relative error of the water quality index of each section simulated by the trained BP neural network model was less than 7%, and all the correlation coefficients exceeded 0.98. The average relative errors of the water quality indices in the verification set were all less than 18%. The prediction accuracy of the proposed model is high and satisfactory, so it can be used for the water quality prediction of Xiaohe River Basin.