Accurate information about electricity generation and consumption is crucial to power system modelling. Several institutions publish such data: for European countries these include the association of system operators ENTSO-E, the EU body Eurostat, and the International Energy Agency; for Germany they comprise the sector organisation BDEW, the federal statistical office Statistisches Bundesamt, the working group AG Energiebilanzen, and the four transmission system operators. This paper compares the terminology, methodology, and reported data of these sources, finding inconsistencies at all three levels. For example, annual electricity generation from wind and solar power in Germany differs by as much as 10% – 20%, depending on who you ask. ENTSO-E publishes “hourly load”, which is widely used among power system modellers. The data documentation provides a (constant) “representativity factor” that should be used to scale the hourly load values. However, we find that the scaling factor, when derived from ENTSO-E’s own more comprehensive data sources (“monthly consumption”), is neither the one provided, nor is it constant. The deviation is particularly worrying in Germany, where peak electricity demand might be underestimated by up to a quarter, and so we propose a scaling procedure that avoids such bias.