Abstract: In order to establish the flood forecast scheme for Yangcun Station on Brahmaputra River, the flood forecast model based on the support vector machine  was studied. According to the characteristics of flood and the hydrological observation station network in Brahmaputra River Basin, the flood forecast model for Yangcun Station with ν-SVR forecast model was established. The model was calibrated and verified respectively by hydrological data and precipitation data from 1988 to 2004 and 2005 to 2007 in flood seasons. The results show that the flood forecast model with ν-SVR can control the number of support vectors, reduce the complexity of the model, and preserve good forecast accuracy, which can be applied in flood forecasting.