This paper proposes a comprehensive evaluation method of node voltage sag severity based on combination weights and TOPSIS model. The SARFI, average temporarily reduced energy index ASEI and sag severity index SSI are regarded as elements to establish the set of attributes. Entropy weight method and coefficient of variation method are used to calculate combination weights of indexes and the voltage sag severity comprehensive evaluation method based on TOPSIS model is established. The sags severity is determined by comparing size of the relative approximate degree of various nodes distance from the positive ideal solution of TOPSIS model. The model is used to assess the grid voltage sag for a certain city. Model assessment results and the index evaluation results are compared from the aspect of the node classification. The results show that the model can highlight the consistency of index evaluation results and reduce the difference, and the results are objective, accurate and more realistic. This work is supported by National Natural Science Foundation of China (No. 51277069).