The change of runoff is closely related to local economic and social development, as well as regional ecological balance and water management. It is important to research the monthly runoff for better reservoir operation and water allocation. To verify the suitability of monthly runoff prediction method, the data of daily precipitation and runoff from eight hydrological stations in Binjiang River Basin were used to predict monthly runoff based on BP artificial neural network, and the results were compared with those from Runoff Coefficient Method, Xin’anjiang Model and HSPF Model. The study shows that BP artificial neural network performes obvious advantages in predicting runoff, its comprehensive uncertainty factor is 0.91, that is much higher than 0.85 of Runoff Coefficient Method and is fairly equivalent to 0.92 of Xin’anjiang Model and 0.96 of HSPF Model. But, the BP artificial neural network model was easier to operate, at the same time, it had accurate simulation for trend, so there might be a good prospect for promoting. However, the results of BP artificial neural network are generally too large, there is certain space for improving.