Abstract: The sandstone and shale’s p-impedances are seriously overlapped in the A oilfield of the Bohai Bay and thus conventional ways using post-stack seismic attribute and inversion cannot effectively predict sand reservoirs. Here we propose a pre-stack joint inversion technology for reservoir prediction to solve this problem. Firstly, reservoir sensitivity analysis is carried out based on logging data. Then, we obtain a sand reservoir identification factor using coordinate rotation technique and make it a target curve. Finally, we obtain a reservoir identification factor volume of the whole area to describe sand reservoir thickness and lateral distribution through multiple attribute neural network inversion with p-impedance, s-impedance, density, and ratio of p-wave and s-wave data acquired by fine pre-stack simultaneous inversion. Actual application shows that this technique can acquire a good reservoir prediction result, which corresponds well with drilling wells vertically and conforms to geological rule horizontally. The result can play a very important role in subsequent drilling and has a broad application prospect.