Abstract: Aiming at the problem of fuzziness and randomness in the process of emergency disasters and accidents decision-making, this article introduces a method of emergency decision-making based on the cloud model. Firstly, decision indexes are selected according to the target and characteristics of rescue in emergency disasters and accidents, and the set of decision indexes is established. Secondly, the weight values and evaluation values of decision indexes are determined according to decision-makers' linguistic values and the relevant range, and the digital characters of the cloud model are produced by backward cloud generator. Thirdly, the evaluation cloud of each alternative is calculated through the arithmetic rules of cloud. Finally, the similarity between each alternative's evaluation cloud and standard evaluation cloud is compared to obtain the comment grade of each alternative, and the best alternative is selected. Case studies show that the method could take full advantage of expert group information, and make the conversion between qualitative description and quantitative indication more effective. The result is objective and feasible, which could improve decision-making objectivity and emergency management effectiveness.