The distribution of voltage sag amplitude and duration is the main characteristic quantity of voltage sag, and it is also an important basis for sensitive load loss estimation and treatment scheme selection. However, the existing stochastic evaluation methods often have shortcomings of needing a lot of calculation and having low accuracy. This makes it difficult to apply the obtained characteristic value of voltage sag in practice. In view of the characteristics of distribution network with many nodes and obvious load variation, the scenario method is applied to the voltage sag characteristic evaluation. First, the K-medoids clustering method is used to classify the node types and select the typical nodes. Secondly, a multidimensional scene dimension reduction of the initial operation state of the distribution network is realized through the node weighted network equivalence, and then the short-circuit fault probability distribution curve and network impedance conversion are combined to propose a generation method of a typical voltage sag scene set. Finally, taking a distribution network structure as an example, the characteristic value of voltage sag at sensitive load points in the power supply area is calculated to verify the effectiveness and feasibility of the method. This work is supported by the Science and Technology Project of State Grid Corporation of China (No. 52153218000B) and the Key Program of National Natural Science Foundation of China (No. 51637007).