Accurate hydrological forecasting is an important non-engineering measure in flood disaster relief. Hydrologic models are the most useful tool for hydrological forecasting. The BP neural network model was improved by introducing LM algorithm, together with the Xin'anjiang Model, their applications for daily flow simulating and forecasting to the Futun River of the Min River were compared. The results showed that, both hydrologic models reached the accuracy requirements of hydrological forecasting with over 90% of hydrological forcasting qualified rate. The Xin'anjiang model performed better for the wet years while the improved BP model was better in simulating accuracy than the Xin'anjiang model. Both models were applicable to the hydrological forecasting of Min River.