Nitrogen flows in agroforestry systems can be quantified by applying excess 15N to one pool or part of the system and subsequently measuring the quantity of 15N in other pools. Accurate quantifications depend on accurate determination of the mass, percentage N, and percentage 15N enrichment of each pool and past studies have mainly used physically isolated subplots to reduce variability. We thus assessed the within-plot and within-plant variability of 15N recovery by maize and by two hedgerow tree species, Gliricidia sepium and Peltophorum dasyrrachis, following applications of 15N-enriched materials to unbounded plots. We also assessed the potential for estimating total plant 15N recovery from a reduced set of samples, and for estimating N content allometrically from simple measurements. 15N uptake beyond 1 m radius was negligible in maize but recommended minimum inter-plot spacing for Gliricidia sepium and Peltophorum dasyrrachis was 8 m. Within-plant variation was also greater in trees than in maize. Calculating recovery from all crop plant components is not always necessary; the recovery of 15N in maize grain predicted 98.9␘f variation in total plant 15N recovery. 89–95ccurate estimates of biomass and N content of maize plants and of regrowing tree shoots can be obtained non-destructively from height or basal shoot diameter respectively. 15N tracing techniques are potentially very useful for studies of competition and complementarity with respect to nitrogen uptake in agroforestry systems but they require unbounded plots and hence particular care in design and sampling procedures.