Abstract: Compared with the Jiyang Depression, the Dongpu Sag of the Bohaiwan Basin has the basic conditions of forming buried hill hydrocarbon reservoirs. Under the control of key wells, the developing conditions of source rocks, reservoirs and caprocks were extracted as input parameters, the grade of hydrocarbon shows was used as indicator parameters, and a neural network mathematical model was constructed to evaluate the hydrocarbon pool-forming conditions of the Upper Paleozoic buried hills in various key wells of the Dongpu Sag. By subdividing the distribution areas of the three types of input evaluation parameters, the pool-forming conditions of the Upper Paleozoic buried hills in the Dongpu Sag were predicted on the plane, which showed that the central uplift zone had the best prospects. Well Wengu-2 achieving industrial productivity for the first time in the Upper Paleozoic buried hills in the Dongpu Sag proved the predicting accuracy of the model. The characteristics of the Wengu-2 hydrocarbon pool indicated that the coarse and thick sandstone above coal measures had lower depth, and should be the main target layers of exploration in the Upper Paleozoic.