In this paper, back-propagation network of the artificial neural network has been employed to establish the frame work of flood forecasting model. The data collected from Wu-Shi Basin has been used to proceed the training of learning network and obtain the model parameters thereby. Further, the recalling process of flood forecasting model was often performed in order to forecast the one step ahead outflow of downstream gauging. The results obtained model verification are satisfactory. The coefficient of efficiency was as high as 0.79, and the error in flood volume was less than 0.13.