Purpose : The Life Cycle Assessment (LCA) has been applied in the construction sector since the 1990s and is now more and more embedded in European public policies, e.g., for Environmental Product Declaration regulation or for building labeling schemes. As far as the authors know, these initiatives mainly rely on background impact data of building products provided by different databases’ providers. The new product-specific and company-specific EPD data allow having more than one data for describing a building material. But are these new databases really displaying similar LCA results compared to generic databases? Does it depend on which impact category (e.g., global warming, acidification, toxicity) is considered? Methods : To answer these research questions, this paper assesses numerical and methodological differences of two existing LCA databases for building LCAs: the ecoinvent generic database and one Environmental Product Declaration (EPD) database developed in France. After reviewing the main assumptions of these databases, numerical values of environmental impact are compared for 28 building materials using Life Cycle Impact Assessment (LCIA) indicators of the EN 15804 standard calculated based on cradle-to-gate ecoinvent and EPD Life Cycle Inventories (LCI). Results and discussion : Global results at the database level indicate deviations of different magnitudes depending on the LCIA indicators and the building materials. While indicators correlated to fossil fuel consumption, such as the ADP, the GWP, and the primary energy demand, exhibit a small deviation (approximately 25 %), other indicators, such as the photochemical ozone formation (POCP), radioactive waste, and ADP elements, are found to be more variable between EPD and generic data (sometimes by more than 100 %). Three indicators are found to be systematically different between EPD and generic data (i.e., the EPD value being either higher or lower for all materials). Similarly, five building materials show systematic differences for all LCIA indicators. Specific deviations for one indicator and one material are also reported. The application of the two databases on three building LCA case studies (brick, reinforced concrete, and timber frame structures) identifies deviations due to the most influential materials. Conclusions : Current generic and EPD databases can present very different values at the database scale which depend on the type of environmental indicator. For building LCA results, the situation is different as generally speaking a limited number of materials controlled the impacts. Finally, recommendations are presented for each environmental indicator to improve the consistency of the building assessment from generic to product- and country-specific information.