The invention discloses a method for predicting the gray-model electric quantity based on day feature vector optimization. According to the invention, with a gray equal-dimensional successive-fillingcomplement model as a main prediction model, daily electric quantity data on the latest ten days are predicted by using a daily feature vector method; and the daily electric quantity data are combinedwith the current-month prediction data to obtain high-accuracy monthly electric quantity prediction results of the current month and the next month. The method disclosed by the invention is simple and is easy to implement; and the effect is obvious. The method can be used as a main method for industry-classified electric quantity prediction and an auxiliary correction method for electric quantityprediction of a power system.