Shallow lakes typically can be in one of two contrasting states: a clear state with submerged macrophytes or a turbid state dominated by phytoplankton. Eutrophication may cause a switch from the clear to the turbid state, if the phosphorus loading exceeds a critical value. Recovery of the clear state is difficult as the critical loading for the switch back during oligotrophication is often lower. A system of interacting ecological processes makes both states stabilize themselves causing the observed hysteresis. The ecosystem of shallow lakes is analysed with PCLake, a dynamic model of nutrient cycling and biota – including phytoplankton, macrophytes and a simplified food web. The model was used to calculate the switchpoints in terms of critical phosphorus loading levels for a number of lake types. It turned out that the predicted critical phosphorus loadings differ per lake type, e.g. they decrease with lake area, mean depth and retention time, increase with relative marsh area and fishing intensity, and differ per sediment type. The findings were grossly comparable with empirical evidence. These outcomes were also used to build a metamodel. The results may be useful for lake management, by comparing the critical loadings for a given lake with the actual loading. If the actual loading clearly exceeds the upper switchpoint, nutrient reduction measures are recommended. If the loading approaches the upper switchpoint, or is in the intermediate range, a manager could try to increase the critical loading values of the lake, e.g. by hydromorphological measures. If the loading is well between the two switchpoints, an alternative is to force a switch by direct food web management.