Abstract. An atmospheric Lagrangian experiment for regional CO 2 budgeting with aircraft measurements took place during the CarboEurope Regional Experiment Strategy campaign (CERES) in south-west France, in June 2005. The atmospheric CO 2 aircraft measurements taken upstream and downstream of an active and homogeneous pine forest revealed a CO 2 depletion in the same air mass, using a Lagrangian strategy. This field experiment was analyzed with a meteorological meso-scale model interactively coupled with a surface scheme, with plant assimilation, ecosystem respiration, anthropogenic CO 2 emissions and sea fluxes. First, the model was carefully validated against observations made close to the surface and in the atmospheric boundary layer. Then, the carbon budget was evaluated using the numerous CERES observations, by upscaling the surface fluxes observations, and using the modeling results, in order to estimate the relative contribution of each physical process.A good agreement is found between the two methods which use the same vegetation map: the estimation of the regional CO 2 surface flux by the Eulerian meso-scale model budget is close to the budget deduced from the upscaling of the observed surface fluxes, and found a budget between −9.4 and −12.1 µmol.m −2 .s −1 , depending on the size of the considered area. Nevertheless, the associated uncertainties are rather large for the upscaling method and reach 50%. A third method, using Lagrangian observations of CO 2Correspondence to: C. Sarrat (claire.sarrat@cnrm.meteo.fr) estimates a regional CO 2 budget a few different and more scattered, (−16.8 µmol.m −2 .s −1 for the small sub-domain and −8.6 µmol.m −2 .s −1 for the larger one). For this budgeting method, we estimate a mean of 31% error, mainly arising from the time of integration between the two measurements of the Lagrangian experiment. The paper describes in details the three methods to assess the regional CO 2 budget and the associated errors.