Abstract: Rockburst is a dynamic breakage phenomenon existing in rock excavation in high in-situ stress zone, and its mechanism is complicated and the occurrence is influenced by many factors. It is almost impossible to make the prediction of its intensity through any single evaluating factor. Based on the idea of information fusion, a methodology is proposed for predicting the rockburst using the evidence theory, which can reflect the comprehensive influences of different factors. Three indices related to the occurrence condition of rockburst are taken into account as evidences in the proposed method, including the ratio of the maximum tangential stresses on cavern boundaries to the uniaxial compressive strength of rock, the ratio of the uniaxial compressive strength to the uniaxial tensile strength of rock and the elastic energy index of rock. The basic probability assignment functions are objectively constructed using the rough set theory. The predicted results of 12 practical cases show that the evidence theory method has relatively high accuracy. Finally, the model is applied to Cangling tunnel and the exploratory tunnel of Jinping II Hydropower Station. The results agree well with the field situations, which again illustrates the practicability of the proposed method.