Abstract: A hydrological forecast model is the core of real-time forecast and warning system. Based on the PDM(Probability Distributed Model)， the probability hydrological forecast model for Feilaixia reservoir in the Beijiang River basin is developed in this paper. The Beijiang River basin is divided into 18 sub-basins. The PDM of each sub-basin is established. Parameters calibration of the whole river basin and the sub-basins is performed by using historical rainfall and flow data. Taking Hengshi hydrological station as a reference station， the developed model is tested and verified by using historical measured flood data from 52 times of flood records. The analysis results show that the developed model is of high accuracy with the qualified rate of flood peak forecasting being 99% and reaching the first precision grade level. The model structure and parameter values of the PDM are of rationality and applicability in Beijiang River basin.After practical application it shows that the model, having a high precision in forecasting, can provide a new reference for hydrological forecasting approach in our country.