dry matter accumulation and increased '^N fertilizer uptake, but less nitrogen fixed on the mineral upland site than on the peat soil.Whether ''N was given as ammonium or nitrate made little difference in these experiments.Grass and clover production and nitrogen cycling were compared in 1983 and 1984 at three sites: an upland peaty gley and upland and lowland brown earths. The clover varieties Olwen and SI84 were compared in 1983 and SIOO and S184 in 1984. Ammonium and nitrate sources of '^N were used to measure nitrogen recovery from fertilizer and soil, nitrogen fixation and nitrogen transfer from clover to grass. Acetylene reduction was measured once, in 1983, but isotope dilution was used in both years.Olwen clover produced more dry matter and took up more '^N than SI84. Olwen fixed more nitrogen than SI84 over the whole season, as measured by '^N isotope dilution. Companion grass took up more soil nitrogen when growing with SI84 than with Olwen. The clover variety SIOO, tested at the lowland site in 1984, caused no significant variations in dry matter accumulation or N fixation.In the dry 1984 season, grass dry matter accumulation and "N uptake were less than in 1983, and plants actually lost total nitrogen from their roots to the soil. Fixation rate varied more in 1984 than in 1983. Nitrogen transfer from clover to grass was detected by isotope ratio differences in 1983, and by total N differences in 1984.Site differences were dominated by the greater dry matter accumulation of Olwen in the lowland in 1983, but in that year there was also increased