Triticum aestivum L. (cv. Gutha), a short-season wheat, was grown to maturity in large monoliths of duplex soil (sand over sandy-clay) in a daylight phytotron mimicking field conditions. Either 15N-labelled ammonium sulphate ((NH4)2SO4) or urea was banded into the soil at a rate of 30 kg N ha-1: even though roots were about 20% heavier when grown in the presence of (NH4)2SO4 for 86 d (P < 0.05), above-ground mass was not affected by the source of nitrogen. At four times through crop development up to grain-filling (50, 56, 70 and 86 d after sowing) shoots were labelled heavily with 14CO2 with two purposes. First, to trace 'instantaneous' assimilate movement over 24 h, revealing relative sink strengths throughout plants. This, in turn, allowed precise measurements of live root mass and the proportion of recent photoassimilates deposited in the rhizosphere. Although root systems were sparse, even in surface soil layers, they were strong sinks for photoassimilates early in development (0-50 d), supporting the conversion of inorganic applied nitrogen (N) to soil organic forms. In the presence of roots, up to 28% of 15N was immobilised, whereas only 12% of labelled ammonium sulphate was immobilised in unplanted plots in spite of a favourable moisture status in both treatments. The effect of plants on rates of 15N transformation is ascribed to recently imported photoassimilates sustaining rhizosphere metabolism. Not more than 15% of recently fixed carbon imported by roots was recovered from the rhizoplane, suggesting that a highly localised microbial biomass supported vigorous immobilisation of soil N. Thus, more than twice as much applied N was destined for soil organic fractions as for root material. By these processes, root- and soil-immobilised N become substantial stores of applied N and together with shoot N accounted for all the applied N under dryland conditions.