The micrometeorological mass balance method has been assessed with a view to its use in the determination of ammonia (NH,) loss from grazed swards. The method involves the measurement of wind speed and the concentration of NH3 in air at different heights above the sward at its windward boundary and a position leeward of the grazed area. The flux of NH3 is calculated from these measurements and a continuous record of wind direction. Quantitative recovery of NH3 from air flows up to 10 litre min-' was achieved using a simple trap containing dilute orthophosphoric acid and a gas dispersion tube. Wind speed measured at a height of 0.25 m at six different positions above a previously grazed sward varied by <2%, The NH3 concentration at the same positions was affected by the distance between the windward edge of the sward and the sampling position (i.e. the fetch) and by heterogeneity in the distribution of sources of NH3 (urine-or dung-affected areas). However, the normalised horizontal flux through 0.25 m (wind speedxNH3 concentration/fetch) varied by less than +8.1% probably due to mixing through fluctuations in wind direction (approximately k30" on the mean) as air passed over the sources of NH3 during each measurement period. Plots of wind speed or NH3 concentration versus logarithm of height indicated that each approximated a linear relationship. This facilitated the calculation of the NH3 flux per unit land area and reduced the total error to about 10%. The total loss of NH3 derived by summing losses during individual sampling intervals of 2 to 18 h within a 24 h period was essentially the same as that derived by averaging wind speeds and NH3 concentrations measured continuously over the same 24 h period. The flux of NH3 from a ryegrass sward grazed by yearling steers ranged from 0.01 to 0.14 kg N ha-' h-' during a 2 day grazing period within a 28 day rotation and during the 5 days following removal of animals. A pronounced diurnal variation was observed in the flux of NH3, the maximum occurring between 13.00 and 20.00 hours on each day. Rainfall and low rates of evapotranspiration reduced the flux to <0.02 kg N ha-' h-'. The total loss of NH3 during 28 days was 20.7 kg N ha-'.