Abstract: The thin interbedded tight sandstone in the Bonan oilfield is a special kind of reservoirs. Multi-phase fans stacked each other with large reserves, but the utilization of reserve remains low in efficiency. Fine description is difficult to achieve since the large buried depth, low vertical resolution and fast lateral variations. To solve the problems mentioned above, joint interpretation technology for multi-sandbodies is used by the authors to establish the fine isochronous stratigraphic framework, and extract and analyze seismic attributes under the guidance of geological sedimentary model. Artificial neural network technology was adopted to establish the non-linear relationship between seismic attributes and reservoir thickness, and to carry out the quantitative prediction of thin interbedded sand body. The research results have successfully applied to the prediction of the thin interbedded tight reservoir in the 176 Block of the Bonan Oil field for well positioning.