Abstract: A fuzzy mode assessment method based on fuzzy evaluation is proposed. It is used to solve the problem that the evaluation results are greatly affected by the subjective factors of the assessors and the weights of the risk factors cannot be reasonably distributed and sometimes it is hard to sort the results of the assessment in traditional risk priority number(RPN). Building the evaluation framework through the "factor set" and "evaluation set". Then use the "trapezoidal fuzzy number" to convert the evaluation result of the evaluator into fuzzy probability. Getting the "fuzzy evaluation matrix" by combining the results of each evaluation and performing "averaged", "anti-fuzzification" and "normalized" processing. Constructing a "weight vector" of risk factors by Analytic Hierarchy Process (AHP). The hazard vector of each failure mode is calculated by the "fuzzy evaluation matrix" and "weight vector". Sort the hazard vector after defuzzification. Extract critical failure modes based on the degree of hazard. Improving the system reliability by reducing the incidence of these failure modes.