Abstract: Prediction of rock burst tendency is one of the basic tasks of disaster prevention and reduction in underground engineering. In view of the existing mathematical models, there are some shortages in the necessary inspection process determining the comprehensive weight of the rock burst index, and there is low accuracy for predicting the rock burst. An improved cloud model for predicting rock bursts is proposed. The improved model is fused with the index weight through cloud atomization conditions in the cloud theory. And the comprehensive weight of the index after the test is obtained. By modifying the standard cloud model and generating the hierarchical integrated cloud by comprehensive cloud algorithm, the deficiencies that the standard cloud model is too sensitive to the average grade range are made up. A number of rock burst examples are used to verify the proposed model. The results show that a more reasonable comprehensive weight can be obtained using the weight fusion method based on the cloud theory. The improved model can be well applied to the prediction of rock burst tendency.