Fluxes of CO2 were calculated by the gradient method from concentration differences, measured in the surface roughness layer above a Douglas fir stand in the Netherlands during a full year (1989). The annual course of the CO2 flux density clearly showed the influence of temperature and incoming radiation on stand assimilation. Monthly mean fluxes were directed towards the stand throughout the year and ranged from 40 kg CO2 ha-1 day-1 in winter to 220 kg CO2 ha-1 day-1 in summer. Results for 1989 were compared with CO2 uptake by the stand, as estimated by a forest growth model, after correction for the soil respiration flux calculated by the model. For the summer months the simulated CO2 uptake agreed with the measurements within the range of uncertainty. During winter-time, however, the model predicted a net loss of CO2 to the atmosphere, whereas the measurements still indicated uptake of CO2 by the ecosystem, in spite of the lower temperature and radiation levels. The difference in the winter months may partially be explained by the uncertainty in the annual course of the soil respiration flux estimated by the model. More probably, however, advection from adjacent deciduous stands and the use of the gradient method beyond its theoretical restraints, may have led to large systematic errors in the fluxes during winter-time.