Gas emission quantity is a crucial factor in the productive process of coal mine. However, because of the complexity of determination, the measuring process is time consuming with a series of norms and manipulations. In our study, we aimed at using the Artificial Neural Networks (ANN) with known experimental data to predict the gas emission quantity. We took seam gas content, embedding depth of coal seam, coal bed thickness, coal bed pitch, working thickness, the length of working face, advancing speed, recovery ratio, gas emission quantity in adjacent layer, the thickness in adjacent layer, the interlayer distance, lithology of interlayer, mining intensity as the independent variables while the gas emission quantity as the dependent variables. By analyzing 18 data groups using General Regression Neural Network (GRNN) and Multilayer Feedfoward Neural Network (MLFN) methods, we found that GRNN model is the best model for predicting the gas emission quantity, with the RMS error 0.50. Results proved that GRNN model is accurate and robust.