The phosphorus cycle in the ecosystem of the shallow, hypertrophic Loosdrecht lakes (The Netherlands) was simulated by means of the dynamic eutrophication model PCLOOS. The model comprises three algal groups, zooplankton, fish, detritus, zoobenthos, sediment detritus and some inorganic phosphorus fractions. All organic compartments are modelled in two elements, carbon and phosphorus. Within the model system, the phosphorus cycle is considered as completely closed. Carbon and phosphorus are described independently, so that the dynamics of the P/C ratios can be modelled. The model has been partly calibrated by a method based on Bayesian statistics combined with a Range Check procedure.Simulations were carried out for Lake Loosdrecht for the periods before and after the restoration measures in 1984, which reduced the external phosphorus loading to the lake from ca. 2 mgP m-* d-' to lmgPm-*d-l.The model outcome was largely comparable with the measured data. Total phosphorus has slowly decreased from an average 130 PgP l-' to ca. 80 PgP l-', but chlorophyll-a (ca. 15opg1-', summer-averaged) and seston concentrations (8-15 mgC l-') hardly changed since the restoration measures. About two-thirds of the seston consisted of detritus, while the phytoplankton remained dominated by filamentous cyanobacteria. The P/C ratio of the seston decreased from ca. 1.0% to 0.7 %, while the P/C ratios of zooplankton, zoobenthos and fish have remained constant and are much higher. The system showed a delayed response to the decreased phosphorus loading until a new equilibrium was reached in ca. five years. Major reasons for the observed resilience of the lake in responding to the load reduction are the high phosphorus assimilation efficiency of the cyanobacteria and the high internal recycling of phosphorus. A further reduction of nutrient loading, perhaps in combination with additional measures like biomanipulation, will be the most fruitful additional restoration measure.