This paper aims to identify the demand elasticity in energy consumption by Brazilian families. By using an original household-level sample, we are the first to identify the average price and income elasticities for Brazil. We also investigate heterogeneous effects across different quantiles through a Quantile Regression (QR) analysis. Finally, we extend the results of the QR analysis and combine them with aggregated data to assess the effects of two price policies, a cross-subsidy and the recently adopted "tariff flags." The primary data comes from the Household Budget Survey (POF) for two periods, 1998-99 and 2008-13, conducted in the metropolitan area of S5o Paulo. The price and income elasticities range from -0.46 to -0.56, and from 0.20 to 0.32 respectively. The QR results indicate that households' behavioral response to a price increase is non-uniform across groups, and that the reaction is stronger for the lower quantiles. Additionally, "tariff flags" lead to a more significant reduction in consumption for the low-income residences, therefore representing a regressive policy. On the other hand, a cross-subsidy that increases the energy price to ordinary residences by 1% could mean a decrease in electricity prices of between 7.5% and 12.4% to low-income families.