The prediction of reservoir sediment can be conducted by the background of Naodehai reservoir so as to meet the need of reservoir operation.The paper first aanalyzed the characteristics of water and sediment in reservoir so as to determine the main factors which affect the sediment storage volume;it further researched to establish sediment storage prediction model based on BP neural network of hydrological factors, and used historical flood data to train and learn. The results showed that the chosen sediment factor can basically reflect the transfer relationship of the processe of rainfall - sediment - sediment transport in the watershed.The model can be used to forecast sediment storage and guide the decision of real-time regulation of reservoir.