We simulated meteorology and atmospheric CO2 transport over the Netherlands with the mesoscale model RAMS-Leaf3 coupled to the biospheric CO2 flux model 5PM. The results were compared with meteorological and CO2 observations, with emphasis on the tall tower of Cabauw. An analysis of the coupled exchange of energy, moisture and CO2 showed that the surface fluxes in the domain strongly influenced the atmospheric properties. The majority of the variability in the afternoon CO2 mixing ratio in the middle of the domain was determined by biospheric and fossil fuel CO2 fluxes in the limited area domain (640×640 km). Variation of the surface CO2 fluxes, reflecting the uncertainty of the parameters in the CO2 flux model 5PM, resulted in a range of simulated atmospheric CO2 mixing ratios of on average 11.7 ppm in the well-mixed boundary layer. Additionally, we found that observed surface energy fluxes and observed atmospheric temperature and moisture could not be reconciled with the simulations. Including this as an uncertainty in the simulation of surface energy fluxes changed simulated atmospheric vertical mixing and horizontal advection, leading to differences in simulated CO2 of on average 1.7 ppm. This is an important source of uncertainty and should be accounted for to avoid biased calculations of the CO2 mixing ratio, but it does not overwhelm the signal in the CO2 mixing ratio due to the uncertainty range of the surface CO2 fluxes.