To comprehensively consider the randomness and ambiguity in rock burst prediction practice, a cloud model theory is introduced into the method. However, the index weight value of the existing rock burst prediction cloud model needs to be improved in terms of objectivity and accuracy. Therefore, this paper proposes a rock burst cloud model based on a back analysis weighting approach. First, the specific implementation steps of the model are given, and the optimization objective function is established. Next, the ratio of maximum tangential stress of cavern to the compressive strength of rock, the ratio of compressive strength of rock to the tensile strength, and the elastic energy index were selected as judging indexes. Based on 18 rock burst engineering examples, the back analysis of index weights was carried out using MATLAB software. Finally, the newly-built cloud model was applied to the rock burst prediction of the Jiangbian hydropower station and Maluping mine. Results were compared with the predictive results of the cloud model based on a subjective weighting method to test the feasibility and effectiveness of the proposed model. The research showed that the subjective interference factors in the weighting process of the rock burst cloud model based on back analysis weighting approach were few, thus the accuracy of prediction results was high.