The consumption of electricity in Brazil has been widely discussed recently. The energy supply crisis in 2001 forced the Federal Government to take a series of measures trying to fix the actual model. Nowadays, it is understood that the energy sector is going through bad times, making the risk of a new rationg plan be considered. In this context, energy companies are facing the challenge of making more accurate load forecast. Consequently, once this need is inserted into a scenario of unstable economy, these estimates require efficient methods combined with innovative features.The aim of this study is to provide a new approach to this electricity prediction problem. A Fuzzy logic methodology is proposed in order to extract rules from the input variables and provide Brazil s Long-term annual electricity demand forecasts. From a statistical modeling point of view, an identification of dependence and lags structure between the input variables provide support for independent models with annual estimates. The advantage of the fuzzy logic model lies on the ability to mimic the human thinking in an environment of uncertainty and imprecision. In recent literature, the formulation of these types of models has been limited to treating the explanatory variables in the univariate form, or involving only the GDP. This study proposes an extension of this model, starting with the Brazilian population and the additional value of the state GDP by sectors with their variations. Then, the proposed model is compared with the official formulation provided by EPE.