Abstract: The neural network models have been studied and developed to establish the non-lineal imaging between the seismic attribute parameters and reservoir parameters. The neural network models can be used to predicate reservoir parameters of the targets and realize the integration of seismic and well logging data, in order to predicate the spatial distribution of reservoir parameters. With actual well logging and seismic data from one of fields in our country, the simulation with the neural network models is finished. The results show that this method is feasible, and the corresponding thematic maps are made.