Abstract: Based on the extension theory method, a multiple indexes identification model for predicting goaf collapse is established with influential factors such as lithology of direct roof, volume rate of the goaf region, direct roof thickness, the vertical goaf depth from the surface, hydrogeologic and engineering geologic conditions, the inclination angle of coal bed and space superposition layer of the goaf region as the evaluation indexes of the forecasting model. Seventeen groups of data selected from the measured data of the goaf region as training samples are tested by the established model, and the correct rate is 100 %. Furthermore, another seven groups of forecast samples are predicted by the multiple indexes identification model. The results show that with the good discriminant performance and accuracy, the model is a new and highly efficient method for the prediction of goaf collapse.