Based on the fuzzy mathematical theory and the self-organizing characteristics of mirror nerval network, a fuzzy identification method for underground excavation stability was proposed which combines the self-organizing mirror nerval network having highly nonlinear characteristics with fuzzy cluster analysis. A clustering analysis model was established for fuzzy analysis of the self-organizing nerval network for evaluation of the stability of underground excavations. For the first time the model was used to determine the stability of stopes. A deep insight was made into the self-organizing characteristics of the stope stability. Case study shows that the method is a scientific classification and is useful for mode identification of stability of underground excavations.