Based on the decision tree and combined with the meteorological factor s and the highest, lowest and average of electrical load, a model was establi shed to measure the influence of meteorological factors on the load. An effectiv e shortterm forecasting method was put forward to establish the forecasting mo del of dailycharacteristicload decision tree. Daily characteristic load is forecasted according to the model and the dateforecasted information such as wea ther，attributes (date, workday or weekend). The forecasted results show that the proposed method has highautomation and highaccuracy. The method applies the data of meteorological factors and load of Baoding in Hebei Province to train an d forecast，the results show that the new method can largely improve the precisi on of load forecasts.