Abstract: In order to reduce the loss caused by the gas accident on coal mine production.this paper using the neural network theory based on the grey model to predict the amount of gas emission in the coal mine,the gray-RBF network model was built,it Make full use the predict characteristics of "small sample of the grey model,poor information" and the predict characteristics self-learning and adaptive ability of RBF neural network.First,using the grey model to make a preliminary forecast,next,Radial basis function network model predict again to get the predicted value of the gas emission eventually,The training of the radial basis function network model and forecast calculation was completed with the MATLAB software.The prediction error of Grey-RBF neural network model are 0.325 and 0.221 respectively,the prediction error of gray model are 2.51 and 2.45,the gray-RBF network model prediction has a higher accuracy degree than the single grey model prediction by comparing the prediction results of gas emission from a mine in Anhui Province,therefore,it provides a method of high precision for gas emission prediction in coal mine.