By using the daily maximum power load data and meteorological data from 9 weather stations over Henan Province from 2005 to 2009, and the stepwise regression method, the summer daily maximum power load prediction models are set up.The models for the whole summer (Scheme 1) and each day of a week (Scheme 2) are established, respectively.The daily maximum power load forecasts for summer 2010 are made by the models.It is found that the daily maximum power loads in working days (Monday to Friday) show different inter annual variation features with weekends (Saturday and Sunday); the daily maximum power load is closely connected with both air temperature and hot index, and can be predicted well by the two schemes.The average prediction errors with the two models for summer 2010 are less than 3%, and the correlation coefficients are large than 09, which shows their good forecasting ability.