Abstract: Reliable and effective medium- and long-term load forecasting is an important basis for power generation and transportation. To achieve this goal, an exponential decomposition analysis based load forecasting method is proposed in this paper, identifying and sorting the key effect factors. Combining the variation trends of the identified key effect factors as well as the social economy, multiple simulation scenarios can be generated for medium- and long-term load forecasting. The historical load and socio-economic data of Beijing city from 1985 to 2014 are selected as the input, and the key effect factors for the load growth are demonstrated. Finally, this paper ends with the load forecasting of Beijing city for the 13th Five-Year Period.