The aim of this study was to understand why two closed dynamic systems with a very similar design gave large differences in soil CO2 efflux measurements (PP systems and LI-COR). Both in the field (forest beech stand) and in the laboratory, the PPsystems gave higher estimations of soil CO2 efflux than the LI-COR system (ranging from 30% to 50%). The difference in wind speed occurring within the soil respiration chambers (0.9 m s(-1) within the SRC-1 and 0.4 m s(-)1 within the LI-6000-09 chambers) may account for the discrepancy between the two systems. An excessive air movement inside the respiration chamber is thought to disrupt the high laminar boundary layer over the forest floor. This would promote an exhaust of the CO2 accumulated into the upper soil layers into the chamber and a lateral diffusion of CO2 in the soil towards the respiration chamber. The discrepancy between the two systems was reduced (i) by decreasing fan speed within the SRC-1, (ii) by increasing wind speed over the soil surface outside the respiration chamber, or (iii) by using an artificial soil design without high CO2 concentration in soil pores. We show that wind speed is an important component of soil CO2 diffusion which must be taken into account when measuring soil CO2 efflux, even on very fine textured soil like silt-loam soil. Proper measurement can be achieved by maintaining wind speed inside the chamber below 0.4 m s(-1) since low wind speed conditions predominate under forest canopies. However, more accurate measurements will be obtained by regulating wind speeds within the chamber at a velocity representative of the wind speed recorded simultaneously at the floor surface.