In order to solve the difficult-to-prediction problem of emergency event evolutionary scenario, the decision paradigm based on similar case analysis is used to propose a method for generating severity degree of emergency events. Firstly, the standardized indexes are used to input the target case for finding similar cases, and filter out several similar cases as reference cases. The severity degree of reference cases is used to acquire the severity degree of the target event. Large-scale group emergency decision-making process is initiated when the average evolution level exceeds level 2. Then, aiming at the problem of the consistency interval existing in a hesitant fuzzy preference matrix, this paper proposes a new method considering the average value and steady state of interval consistency to distribute experts'weight. Finally, an example is given to illustrate the effectiveness of the proposed method.