Published field data for lupins grown in a deep sandy soil in the wheatbelt of south-western Australia were used to test the predictive ability of a model of three-dimensional root growth. The model has the capacity to simulate the growth of individual root sections in response to the supply and demand for water and nitrate. N mineralisation was not modelled explicitly, but was accounted for through the use of a seasonally variable mineralisation input derived from the field data. Simulated nitrogen and water contents and root length densities in the soil profile agreed well with observed profiles, although all were slightly under-predicted. A sensitivity analysis revealed that model predictions were most sensitive to the drained upper limit values (v/v) and the mineralisation rates (mugN m(-3) s(-1)) incorporated as external inputs to the model, along with the unit rate of N-2 fixation (mol nodule(-1) s(-1)) and unit root growth rates (mum mol(-1) s(-1)) which are physiological parameters previously calibrated for lupins, The amount of nitrate leached was predicted well. Spatial plots of nitrate leaching were a close inverse of the root length density plots, with the highest nitrate leaching below the inter-plant zones, and the least nitrate leaching directly below each plant. These results suggest that the root distribution of a legume species such as lupin can have an effect on the leaching of nitrate to depth. It may thus be possible to reduce the total amount of nitrate leached under lupin crops by investigating factors such as the spatial deployment of roots, planting densities and intercropping.