A leaf-feeding technique for in situ N-15-labelling of intact soil-pasture plant systems was assessed, using subterranean clover (Trifolium subterraneum L.) and serradella (Ornithopus compressus L.) grown under glasshouse conditions. Total recoveries of fed N-15 were 87-100% following leaf-feeding of plants at flowering but were lower (74-84%) following the feed at the vegetative stage. Below-ground recovery of fed N-15 ranged from 7 to 26%, with serradella partitioning a greater proportion of labelled N below ground than subterranean clover. Additionally, plants of both species fed at the vegetative stage accumulated a greater proportion of the N-15 label below ground than did those fed at flowering. Dry sampling procedures, which utilised freeze-drying, enabled fractionation of the below-ground portion of the system into 'clean' nodulated macro-roots with no adhering soil, residual uncleaned root, rhizosphere, and bulk soil. Calculated specific enrichment for the 'clean' roots at different depths demonstrated a relatively uniform distribution of N-15 label in the subterranean clover roots: whereas the presence of large indeterminate nodules in the crown region of serradella roots contributed to apparent uneven distribution of label. Approximately half of the N in the residual fraction of both species consisted of labelled material, postulated to be mostly fine root. Additionally, 5-20% of the rhizosphere N and 0.5-3% of the N in Sulk soil was legume root-derived, with some N-15 detected in the extractable total soluble N and microbial N pools. Rhizodeposition of N represented approximately 10% of total plant N and 17-24% of total below-ground N for subterranean clover, whereas values for serradella were 20 and 34-37%, respectively. Estimated total below-ground N of subterranean clover reached a maximum value of 177 mg N/plant at 98 days after sowing, which corresponded with a peak shoot N of 243 mg N. Maximum below-ground N for serradella attained 196 mg N/plant 84 days after sowing with a corresponding shoot biomass of 225 mg N. There was a decline in the total below-ground N of serradella at maturity. Overall, recovered clean root N represented 30-62% of estimated total below-ground N; so it was concluded that standard root recovery procedures might be likely to underestimate severely the total below-ground N accretion and N turnover by legumes. The implications of these results for field estimation of total legume N yield, biological N fixation, and the N benefit from legumes in rotations are discussed.