Aiming at he prediction of oilfield development indexes, this paper puts forward a prediction model(IPSO-GNN) combining Gray Neural Network(GNN) and Improved Particle Swarm Optimization algorithm(IPSO),to make up for the shortage of the GNN and guarantee the forecast accuracy, it optimized the GNN parameters through the IPSO. With the prediction of moisture content in oilfield development indexes as an example, the simulation results show that the prediction accuracy of the combination prediction model it higher than gray prediction model, the gray neural network and the BP neural network model. It also shows the feasibility and effectiveness of the method.