On the basis of the analyses of twenty year data observed in Jiuyuangou watershed on the Loess Plateau,a model of sediment yield of watershed under single rainfall is propounded through the application of back propagition artificial neural network.The network model is trained and predicted by input rainfall,average rainfall intensity,runoff depth and flood-peak modulus.The predicted results show that the network model has good precision and accurately reflects the laws of nonlinear sediment yield of watershed under single rainfall.It provides a new foundation to establish the predicted model of sediment yield of watershed under single rainfall on the Losses Plateau.