Abstract Using a unique dataset of 256,000 residential property sales for Portugal, this paper reveals a clear sales premium for energy efficiency, which is more pronounced for apartments (13%) than for houses (5 to 6%). Price premiums tend to increase from 2009 to 2013, a period in which the Portuguese housing market was depressed. Quantile regression reveals that, when compared with the rest of the sample, the group of the most energy efficient properties receives a statistically relevant price premium, that is mostly stable across the entire spectrum of the conditional price distribution. Cross-country comparisons suggest that energy efficiency price premiums are higher than those found for central and northern European markets. Illustrations on the effect of data issues in hedonic regression models are provided. They show how the use of appraisal prices and explanatory variables with measurement errors may seriously bias energy efficiency partial effect estimates. In contrast, the omission of variables associated with the quality of the properties has not produced relevant distortions. Finally, it became apparent that the use of smaller datasets would have produced similar results, as no significance inflation was produced by the Portuguese large scale dataset.