Abstract: Grain production is an important part of the national economy, and grain yield is significant for guaranteeing grain security in China. In order to improve the scientific and accurate prediction of grain yield, based on the analysis of the existing prediction methods, the grey theory and neural network were combined organically, the main and objective factor indexes were determined from many influencing factors on grain yield by grey correlation analysis. These indexes were applied in the prediction of grain yield with the artificial neural network. The results showed that the maximum error of the artificial neural network prediction was 1.21%, the average error was 0.63%, and the prediction accuracy was relatively high. This study provides a scientific and effective forecasting method for grain yield.