In trials of a mass balance method for measuring methane (CH 4 ) emissions, sonic anemometers and an open-path laser were used to measure the transport of CH 4 released from a ground-level source across a downwind face 50m long and 6m high. Release rates matched emissions expected from dairy herds of 2 to 40 cows. The long laser path permitted inferences from measurements in only two planes, one upwind and one downwind, while the fast-response instruments allowed calculation of instantaneous horizontal fluxes rather than fluxes calculated from mean wind speeds and mean concentrations. The detection limit of the lasers was 0.02ppmv, with the separation between the transmitters and reflectors being about 50m. The main conclusions from the 23 trials were: (1) Emissions calculated from mean wind speeds and concentrations overestimated the true emissions calculated from instantaneous measurements by 5%. (2) Because of small changes in methane concentration, the minimum sample size in animal trials would be 10 dairy cows, producing about 40mg CH 4 s −1 . (3) For release rates greater than 40mg CH 4 s −1 and with sufficient replication, the technique could detect a change in production rate of 9% (P<=0.05). (4) Attention to perceived weaknesses in the present technique should help towards detecting changes of 5%.