We analyzed measurements of net ecosystem exchange of CO2 (NEE) over 15 European forests (the EuroFlux data set) to investigate which climate and forest characteristics explain temporal and intersite variations in NEE and its components, gross primary production (GPP) and respiration (R). Informed stepwise regression was used to derive a parameter-efficient, empirical model that was consistent with process knowledge. The resulting model required seven site-specific parameters to describe flux behavior at different temporal scales as a function of radiation, temperature, and air humidity. The interpretation appeared robust despite method and data uncertainties, although the data set was probably biased toward well-watered boreal and temperate European forests. Radiation, temperature, and leaf area (through forest assimilation capacity) appear to be the main drivers of the observed temporal and intersite variation in gross primary production, ecosystem respiration, and net ecosystem exchange.