The objectives of this thesis are (i) to quantify NH 3 volatilization from grassland, (ii) to gain understanding of the NH3 volatilization processes on grassland and (iii) to study measures how to reduce NH 3 volatilization from grassland as such and on farm scale. Volatilization of NH3 was studied of N fertilized grassland under grazing by dairy cows, of mown N fertilized grassland during regrowth and of cattle slurry application.Grazed pastures on a loamy soil showed large variations in NH 3 emissions between individual cuts. Rainfall and sward management affected NH 3 volatilization rate. There was not much difference in NH 3 emissions between years. At a N fertilizer rate of 550 kg ha -1yr -1, NH 3 losses were 8.5%, 7.7% and 6.9% in 1987, 1988 and 1990, respectively, of the N returned to the sward in urine and faeces. At 250 kg N ha -1yr -1they were 3.1 % and 3.3 % in 1988 and 1990, respectively. Whole season measurements are required to obtain reliable data on NH 3 volatilization. Volatilization of NH 3 was related to N excretion and N application rate. A calculation model was developed to estimate NH 3 volatilization from N application rate. Measurements on the same soil showed that a mown-only sward released NH 3 during a few days after fertilization with calcium nitrate (100 kg N ha -1). Thereafter a net uptake was observed until harvest. On average a net uptake of aerial NH3 was measured of a few kg N ha -1per regrowth period. The measured NH3 fluxes were much greater during the day than during the night. The NH3 compensation point (on average 14 μg NH 3 -N m -3) could be related to grass N content and soil water content. It is suggested that on grazed swards reabsorption of emitted NH 3 from urine and dung spots is important. Because of this micrometeorological techniques are required to determine NH 3 losses from grazed swards.Urea (U) fertilizers may emit NH 3 whereas calcium ammonium nitrate (CAN) fertilizers may emit N 2 O or may be susceptible for nitrate leaching. Agronomic trials in the Netherlands (NL) confirmed the superiority of CAN to urea, whereas in the UK and Ireland U was effective as CAN, especially in the first cut. It was proved that especially the higher amount of rainfall in the UK and Ireland compared to NL caused the observed differences. Optimizing herbage yield, thereby minimizing N losses may be the best approach to decide which fertilizer to use. A calculation model was developed to improve decision analysis whether to use U or CAN. Under prevailing NL conditions for the first and second cut, it is only once every 5 and 7 years profitable to use U instead of CAN.Acidification of cattle slurry with nitric acid resulted in marked reductions in NH 3 emissions. Micrometeorological. experiments showed reductions of on average 85, 72 and 55% for acidified slurry with pH values of 4.5, 5.0 and 6.0, respectively. The measured losses could be related to temperature, slurry pH, potential water evaporation and NH+4concentration. It was possible to calculate which pH would be required to obtain a certain reduction in NH 3 emission.Losses of NH 3 occur during slurry application, housing, slurry storage, grazing, fertilizer application and from crops, in descending order of importance. The main strategies to reduce potential NH 3 loss are lowering the excretion of urea with urine and/or reducing the rate of NH 3 from slurry loss by technical means. However, reducing NH 3 loss may increase other N losses. To prevent this a farm scale approach is required. Modelling showed great scope for reducing NH 3 3 losses and N budget surpluses on dairy farms by improved management.