[en] P>Soil respiration measurements performed with closed dynamic chambers are very sensitive to pressure differences inside and outside the chamber: differences as small as 1 Pa can induce errors that are of the same magnitude as the flux itself. The problem is usually solved by adding a vent to the experimental set-up. However, although this may give acceptable results in most cases, it is not effective at sites that are very exposed to wind. At the CarboEurope-IP agricultural site of Lonzee (Belgium), on bare soil, we used a vent composed of a vertical tube whose upper end was placed between two horizontal plates. Whilst this system worked properly in low-wind conditions, it led to a significant flux over-estimation (up to 300%) under strong wind conditions. We analysed the causes of this error and attributed it to a dynamic pressure effect at the vent, leading to air aspiration from within the chamber. We suggest that this is because the wind at the vent level was not the same as that experienced in the chamber, because of the large vertical wind speed gradient close to the soil. Another vent geometry was then proposed that positioned the vent at the chamber level. This new design was tested on both manual and automatically operated chambers. It was found to be efficient in windy conditions as most of the artificial correlation between soil respiration measurements and wind speed had disappeared.