Abstract: In order to achieve an accurate assessment of the safety operation state of tailings pond, a evaluation model of safety operation state of tailings pond based on cloud-DS evidence theory is proposed. The evaluation index system of safety operation state of tailings pond is determined based on bibliometric method, and the quantitative range of each safety operation state level is divided. The cloud digital characteristics of each index and the degree of certainty of each level are obtained based on the normal distribution cloud model, and constructs into a basic probability assignment matrix, which is transformed into the basic probability assignment of the evaluation index system of safety operation state of tailings pond; an index weight determination method based on game theory is proposed, and the evidence obtained from the basic probability assignment of evaluation index of safety operation state of tailings pond is combined with evidence fusion to obtain the basic probability assignment of different safety operation state of tailings pond. Based on the principle of maximum affiliation, the corresponding safety operation state level is determined as the final evidence fusion evaluation results. The assessment results of the model are verified by the current situation of the tailings pond and the actual safety state of the tailings pond. It shows that the accuracy of the model is high, and it has good applicability for the safety state assessment of tailings pond, which can provide theoretical basis for the formulation of safety management policies related to tailings pond.