Abstract: As to single-event voltage sags, it is still difficult to know how serious they are even after the assessment value is got because there's no reference value. A novel classification method for assessing single-event voltage sags is proposed with the consideration of equipment sensitivity. Firstly, the probability distribution functions of equipment immunity level are fitted according to the severity evaluation results of typical tolerance curves of equipment using the voltage sag energy index. Then the probability of equipment failure caused by single-event voltage sag is calculated, and the comprehensive failure rate is obtained based on the analytic hierarchy process. Further, the sag energy index is modified by constructing a correction function by using the comprehensive failure rate as the input variable. Based on 490 groups of measured sag samples, a large amount of modified sag energy index values are calculated to carry out probability statistics analysis, its probability distribution function is fitted, and several probability quantile values are calculated to be used as the reference value. In the end, the classification levels of voltage sag severity are established by comparing the modified evaluation index with the reference value. It is testified that the proposed method can efficiently reduce the evaluation error of the sag energy index by comparing the results before and after modification of sag energy index through classification evaluation of 100 measured sags samples of power system, and assessment results are more intuitive and reasonable.