The randomness of EV charging and discharging behavior leads to strong fluctuation of running data and large difference of weights among indexes, thus it is difficult for traditional methods to evaluate the grid-connected performance of EV charge and discharge stations comprehensively and accurately. Therefore, a comprehensive evaluation method for grid-connected performance of EV charge and discharge stations is proposed. First, a comprehensive evaluation index system including 5 criteria such as safety, adaptability and stability is proposed. Second, the Apriori algorithm is used to process the correlation between indicators. At the same time, a game idea is introduced to combine the ANP and CRITIC methods to determine the optimal combination weight of indicators. Based on the optimal combination weight and the TOPSIS method, an evaluation model is established to quantitatively analyze the comprehensive performance of grid-connected charge and discharge stations. Finally, the proposed method is verified by using the operating data of an EV charge and discharge station in a city, and compared with the existing evaluation methods. The results show that the proposed evaluation system and evaluation method are effective and superior.