There exist a number of important factors in forecasting processes. The most significant in the case of a multi-dimensional econometric model are the choice of the model and the explanatory variables. This paper focuses on the choice of variables, which plays a crucial role. Among many existing and recognized algorithms for the selection of variables, the following three chosen arbitrarily by the authors are presented: the method of integral capacity by Hellwig, the partial correlation coefficient, and multiple correlation coefficient. For each considered method, a synthetic description explaining its action is given. The choice of the model for making forecasts was treated as less significant because, for each method, the same model is used. Calculations were made using the MRK (Model of Canonical Distribution) model. The synthetic description of the model is also provided. Using real-life data, the analysis demonstrates how the method applied in choosing explanatory variables influences the obtained forecast results concerning the consumption of electric energy on a national scale. In the example calculation, a potential dataset of seven explanatory variables was used. The paper summarizes fourteen different forecasts obtained from three methods of variables selection. The results of model fittings (extinct forecasts) and forecasts until 2030 are presented in the form of tables and charts. Concluding remarks concern long-term forecasts of fundamental quantities related to the Domestic Power Engineering System, which may influence the correctness of these forecasts. Particular attention was paid to the issue of the correctness of long-term forecasts related to energy consumption and peak power. An analysis of the forecast of energy consumption, processed parallel to the determination of quantities assessing the reality of these forecasts, makes it possible to indicate the forecasts endowed with the lowest risk of making excessive errors. For example, it may be possible to consider the determination of annual load levels for which a pre-determined, exact level of variability is given. In situations where only the forecast of electric energy consumption is available, it is possible to develop further analyses such as a comparison of energy consumption for a Domestic Brutto Product, energy consumption for products, or economic branches in selected countries. Yet another indicator in common use is annual energy consumption per capita. Forecasts per capita are among the most exact; therefore, this indicator may be a useful tool for comparison of different forecasts.