Methane (CH4) from ruminants contributes one-third of global agricultural greenhouse gas emissions. Eddy covariance (EC) technique has been extensively used at various flux sites to investigate carbon dioxide exchange of ecosystems. Since the development of fast CH4 analyzers, the instrumentation at many flux sites has been amended for these gases. However, the application of EC over pastures is challenging due to the spatially and temporally uneven distribution of CH4 point sources induced by the grazing animals. We applied EC measurements during one grazing season over a pasture with 20 dairy cows (mean milk yield: 22.7 kg d−1) managed in a rotational grazing system. Individual cow positions were recorded by GPS trackers to attribute fluxes to animal emissions using a footprint model. Methane fluxes with cows in the footprint were up to 2 orders of magnitude higher than ecosystem fluxes without cows. Mean cow emissions of 423 ± 24 g CH4 head−1 d−1 (best estimate from this study) correspond well to animal respiration chamber measurements reported in the literature. However, a systematic effect of the distance between source and EC tower on cow emissions was found, which is attributed to the analytical footprint model used. We show that the EC method allows one to determine CH4 emissions of cows on a pasture if the data evaluation is adjusted for this purpose and if some cow distribution information is available.