Purpose: Fuel demand forecast is a fundamental tool to guide private planning actions and public policies aim to guarantee energy supply. In this paper, different forecasting methods were evaluated to project the consumption of light fuels in Brazil (fuels used by vehicles with an internal combustion engine).

Design: Eight different methods were implemented, besides of ensemble learning technics that combine the different models. The evaluation was carried out based on the forecast error for a forecast horizon of 3, 6 and 12 months.

Findings: The statistical tests performed indicated the superiority of the evaluated models compared to a naive forecasting method. Furthermore, for 12 months forecast, it was found methods that outperform, with statistical significance, the SARIMA method, that is widely used. The results indicate, for all forecast horizon, that is possible to estimate a model whose mean absolute percentage error is less than 3%.

Practical implications: The level of accuracy reached allows the use of these models as tools to assist public and private agents that operate in this market.

Originality: The study seeks to fill a gap in the literature on the Brazilian light fuel market. In addition, the methodological strategy adopted assesses projection models from different areas of knowledge using a robust evaluation procedure.