As the tunnel surrounding rock conditions in Yuandian mine is complex, the current classification scheme is not perfect, and the roadway convergence rate is larger, this paper attempts to explore an accurate, efficient and easy prediction method of roadway surrounding rock stability. Taking Yuandian mine roadway as the engineering research background, this paper uses the BP neural network forecasting principles, and designs a classification scheme of surrounding rock stability of roadway based on BP neural network after an on-site analysis of the factors influencing the stability of tunnel surrounding rock. By using MATLAB7.0BP neural network toolbox, it realizes the algorithm programming and GUI friendly user’s interface. Results show that this method has a high precision of forecasting.