Abstract The EU directive to create Energy Performance Certificates (EPC) for all buildings was implemented in Sweden as a tool to advise building owners on possible improvements and to give energy efficiency visible market value. The Swedish EPCs include measured energy usage. Currently 82% of the buildings have EPCs; this database makes it possible to create overview and to validate models of the building stock in an unprecedented high detail. However, the process of issuing Swedish EPCs has received criticism from real estate agents, real estate owners, Energy Experts, and Boverket, the national agency responsible for EPC data collection. In order to use the EPC data for describing the building stock it is necessary to assess and remediate the data quality. This has been done by merging the EPC data with databases of the Housing and Urban Development office and one of the larger real estate companies in Sweden, Riksbyggen. The Swedish EPC specific area measurement, Atemp, is found to vary according to methods of derivation. The method of estimating Atemp is improved using a stepwise regression model (R2 = 0.979). This method can be applied to subsets of EPCs depending on the intended way of describing the building stock.