Abstract: There is a large uncertainty in the participation of electric vehicle（EV） in demand response, which brings great challenges to the prediction of EV loads. Therefore, an EV day-ahead load forecasting method considering the uncertainty of demand response was proposed. Firstly, an EV automatic demand response strategy was designed by combining two demand response mechanisms: price and incentive. Then, in order to describe the randomness and fuzziness of user’s response to the strategy, a new Sigmoid cloud model was proposed to describe the uncertainty mapping relationship between user’s profitability and response behavior, which is affected by complex factors, and users’ acceptance of the strategy were obtained. Furthermore, EV loads and their probability distributions under different response scenarios were calculated. The simulation results show that, compared with the traditional methods, the proposed Sigmoid cloud quantizes the uncertainty of EV response behavior, the RMSE is reduced by about 60%, and the relative errors of EV load probabilistic prediction results are maintained at 0.9%～9.6%. The prediction model and method are more precise and accurate.