Sugar beet (Beta vulgaris L.) was grown in two different longterm P fertilization experiments on a sandy and a loamy soil. The P supply levels of the soils were "low", "sufficient", and "high", according to the German recommendation scheme. The low P level decreased shoot and storage root yield only on the loam soil, where the recovery of the P-deficient plants after a drought period was slower than at a sufficient P supply. The size of the living root system, as determined by a conventional auger sampling method, peaked at early July and decreased until harvest on the sandy soil without any influence of the P level. On loam, the living root systems were more constant and larger at P shortage. Total root production, as determined by the ingrowth core method, was about 120 km, m(-2) in the well P supplied loam treatments and 200 km m(-2) at P deficiency, which was 3-4 times and 5 times higher than the average size of the living root systems, respectively. Hence, a rapid root renewal took place. On sand, where no P deficiency occurred, total root production was not different between the P supply levels but higher than in the well-supplied loam treatments. Modelling P uptake revealed that this root turnover and the concomitant better exploitation of the soil facilitates P uptake at a low P level in soil, but is of no advantage at a sufficient P supply. The increase of root production at P shortage increased calculated P uptake by 25% compared to a calculation with the "usual" root production at a sufficient supply.