Abstract: Rockburst is one of the main geological disasters during construction in deep and long tunnel. The whole process of 'gestation-occurrence-evolution' is extremely complicated. The prediction of rockburst has become a worldwide difficult problem in underground engineering. It is of great significance to accurately reflect the change law of high ground stress and the influence of geological structure in engineering region, and based on the measured data to predict rockburst classification to ensure the preliminary geological survey and safe construction of tunnels. According to the regional geological data of tunnel, we use the COMSOL Multiphysics and Rhino establish a three-dimensional numerical calculation model. By using the boundary load adjustment method, we inverse the whole area of in-situ stress field. Considering the key factors of rockburst comprehensively, four factors are selected as evaluation indexes. They are the ratio of the uniaxial compressive strength(UCS) of rock to maximum principal stress of the cavern (σ_c/σ_max), the ratio of the maximum tangential stress of the cavern to the UCS of rock (σ_θ/σ_c), the ratio of UCS to tensile strength of rock(σ_c/σ_t), and the elastic strain energy index (W_et). We further use the entropy weight and efficacy coefficient method to determine the weight of indexes and the total efficacy coefficient separately. We construct a rockburst prediction model of deep and long tunnel using the efficiency coefficient method and the ground stress field inversion. The evaluation model is applied to Sangzhuling tunnel and compared with the actual situation to verify the result. The result shows that this method has high accuracy and reliability for rockburst prediction of deep and long tunnel. It provides a new basis for the prediction of rockburst in similar underground engineering.