The relative distributions of tree and crop roots in agroforestry associations may affect the degree of complementarity which can be achieved in their capture of below ground resources. Trees which root more deeply than crops may intercept leaching nitrogen and thus improve nitrogen use efficiency. This hypothesis was tested by injection of small doses of (15NH4)2SO4 at 21.8 atom 5N at different soil depths within established hedgerow intercropping systems on an Ultisol in Lampung, Indonesia. In the top 10 cm of soil in intercrops of maize and trees, root length density (Lrv) of maize was greater than that of Gliricidia sepium trees, which had greater Lrv in this topsoil layer than Peltophorum dasyrrachis trees. Peltophorum trees had a greater proportion of their roots in deeper soil layers than Gliricidia or maize. These vertical root distributions were related to the pattern of recovery of 15N placed at different soil depths; more 15N was recovered by maize and Gliricidia from placements at 5 cm depth than from placements at 45 or 65 cm depth. Peltophorum recovered similar amounts of 15N from placements at each of these depths, and hence had a deeper N uptake distribution than Gliricidia or maize. Differences in tree Lrv across the cropping alley were comparatively small, and there was no significant difference (P<0.05) in the uptake of 15N placed in topsoil at different distances from hedgerows. A greater proportion of the 15N recovered by maize was found in grain following 15N placement at 45 cm or 65 cm depth than following placement at 5 cm depth, reflecting the later arrival of maize roots in these deeper soil layers. Thus trees have an important role in preventing N leaching from subsoil during early crop establishment, although they themselves showed a lag phase in 15N uptake after pruning. Residual 15N enrichment in soil was strongly related to application depth even 406 days after 15N placement, demonstrating the validity of this approach to mapping root activity distributions.