Abstract: To master the building damage degree in mining areas, fuzzy pattern recognition method is used to forecast the damage grade. Firstly, 30 known datum of building damage in mining areas are preprocessed, then a model base including four standard models is established by fuzzy clustering analysis method, according to the underground mining situation, geological conditions and building structure and so on. Afterwards, eight samples have been tested with the standard model base, the closeness between the sample and each of the standard model is calculated, and then the corresponding standard model of each sample is determined based on the principle of choosing nearest. Calculation result shows that the predicted result of the 8 test samples equal to the measured values, therefore they are accurate and reliable. The study shows that using fuzzy recognition method to predict building damage grade is feasible, and it is a new approach of predicting the damage grade of buildings in mining areas.