This work analyzes the impact of grazing on the carbon balance of a grassland grazed by the Belgian Blue breed of cattle. The research was run at the Dorinne terrestrial observatory (DTO). The experimental site is a permanent grassland of ca. 4.2 ha located in the Belgian Condroz (50° 18’ 44’’ N; 4° 58’ 07’’ E; 248 m asl.). Other studies are conducted at the DTO including measurements of methane (CH4) and nitrous oxide fluxes (Dumortier et al., Geophysical Research Abstracts, Vol. 15, EGU2013-2083-1, 2013; Beekkerk van Ruth et al., Geophysical Research Abstracts, Vol. 15, EGU2013-3211, 2013, respectively). Grassland carbon budget (Net Biome Productivity, NBP) was calculated from Net Ecosystem Exchange (NEE) measured by eddy covariance by taking imports and exports of organic C and losses of carbon as CH4 into account (Soussana et al., 2010). After 2 years of measurements (May 2010 - May 2012), the grassland behaved on average as a CO2 source (NEE = 73 ±31 g C m-2 y-1). After inclusion of all the C inputs and outputs the site was closed to equilibrium (NBP = 23 ±34 g C m-2 y-1). To analyze the impact of grazing on CO2 fluxes, we studied the temporal evolution of gross maximal photosynthetic capacity GPPmax and dark respiration Rd (deduced from the response of daytime fluxes to radiation over 5-day windows). We calculated GPPmax and Rd variation between the end and the beginning of grazing or non-grazing periods (∆GPPmax and ∆Rd, respectively). We observed a significant decrease of GPPmax during grazing periods and measured a ∆GPPmax dependence on the average stocking rate. This allows us to quantify the assimilation reduction due to grass consumption by cattle. On the contrary, no Rd decrease was observed during grazing periods. Moreover, we found that cumulated monthly NEE increased significantly with the average stocking rate. In addition, a confinement experiment was carried out in order to analyze livestock contribution to Total Ecosystem Respiration. Each experiment extended over two days: the first day, ...