Abstract: Paleogene Huagang Formation in Xihu Sag is dominated by deltaic deposits, of which channel sand bodies are developed. Accurate prediction of channel sands is key to hydrocarbon exploration in this formation. However, it is very difficult to identify seismic response characteristics of channel sand by conventional seismic prediction methods, due to the similar velocities and slightly different wave impedance between sandstones and mudstones in the study area. This study proposes a new method of channel sands identification and prediction based on reservoir AVO characteristics. Firstly, the response characteristics of the main sand bodies is defined through fine calibration combined with rock physics analysis and forward modeling. Secondly, according to the AVO characteristics of the channel sand body, the sensitive far-offset stacking data is selected, and frequency division fusion technology is used to characterize the channel sand body contour. Finally, the correlation between the sensitive attributes of advantage prestack angle stacking data and the thickness of the sand body is studied and the correlation formula is established to accurately depict the distribution of the channel sand body. Good results have been achieved in practice. This technique provides valuable reference for the exploration and development of channel facies sand bodies in similar areas.