We present a methodology to detect inconsistencies in perhaps the most important data source for measurements of ecosystem–atmosphere carbon, water, and energy fluxes. We expect that the derived consistency flags will be relevant for data users and will help in improving our understanding of and our ability to model ecosystem–climate interactions.