Based on attribute mathematical theory, an attribute recognition model to predict and classify rockbursts was established. Firstly, the main factors of rockburst, such as the maximum tangential stress of cavern walls σθ, uniaxial compressive strength σc, uniaxial tensile strength σt, and the elastic energy index of rockWet, were chosen for the analysis; and three factors, includingσθ/σc，σc/σt andWet, were chosen as the criterion indices for rockburst prediction in the proposed model. Secondly, attribute measurement functions were constructed to compute the attribute measurement of a single index. Thirdly, the index weight was determined by similar weights defined by similar figures. Finally, the possibility and classification of rockburst were recognized by the confidence criterion. A series of underground rock projects were assessed with the proposed model and method to verify the proposed model. The study indicates that the synthetic assessment results agree well with the practical records, and are coherent to those of the fuzzy synthetic evaluation model and the matter-elements model. Moreover, the proposed model was used to predict rockbursts of a hydropower station and Qinling Tunnel. The results are coherent to those of the synthetic evaluation method, such as artificial neural network and distance discriminant analysis method, and others. The research indicates that an attribute recognition model can predict and classify rockbursts in engineering projects deep underground and provides a new method in practice.