This paper uses the direct account method to predict energy consumption at two hierarchical levels: the upper level corresponds to the country, and the lower to industrial production, social sphere, and population. When forecasting at the upper level, we applied the parameters of GDP energy intensity and the GDP forecast with estimation of the energy-saving potential from structural and technological changes. At the lower level, the parameters of energy efficiency were applied according to each group: industrial production – specific consumptions of energy resources and production forecast, social sector – specific consumptions of fuel and energy resources for municipal and other economic services, and forecast of the volume of services provided, population – demographic forecast and predicted volume of heated housing. For the prediction of energy consumption in industrial production, from 51 types of industrial products, 26 most energy-intensive ones were selected, which, in total, consume at least 70% of energy resources and have high specific consumptions of energy resources for at least one type of them. For each of 26 types of products, we chose measures of energy savings and estimated the potential for energy savings up to 2040. We calculated the forecasts for fuel, heat, and electricity consumption by 26 types of products, the forecasts of consumption of total fuel and energy resources up to 2040 for energy-intensive products, population, and social sphere. A comparison of the correspondence between obtained forecasts at the upper and lower levels for certain types of energy resources is made. We conclude that it is necessary to combine several methods in forecasting energy consumption for different types of energy resources, for example, normative and direct account.