Abstract: The best method to distinguish the components of the variations in daily electrical loads and the main factors that cause daily maxima, and the optimum number of forecast models was explored. Based on daily electric load, temperature, relative humidity, rainfall, and wind speed data from 2005 to 2010 in Beijing, the characteristics of the daily electric load maxima were analyzed and the methods to distinguish the variation components were discussed. Based on the correlation between electric load and meteorological parameters, the maximum electric load of the long summer days was also analyzed. The method of stepwise regression was used to model the variations in daily electric load maxima from June to August during 2005-2009. The results showed that the daily electric load increased linearly from 2005 to 2010 with a significant weekly effect during summer. Separation methods based on annual daily variation trends are the best. For the same day, the correlation coefficient between electric load and average air temperature is high, which also applies to the variations of the daily electric load for the day before. Seven forecast models that take into account the daily air temperature and electric load for the same day and the day before are better than two or four models that consider the whole week.