1. Shallow lake ecosystems are normally dominated by submerged and emergent plants. Biological stabilising mechanisms help preserve this dominance. The systems may switch to dominance by phytoplankton, however, with loss of submerged plants. This process usually takes place against a background of increasing nutrient loadings but also requires additional switch mechanisms, which damage the plants or interfere with their stabilising mechanisms. 2. The extent to which the details or even major features of this general model may change with geographical location are not clear. Manipulation of the fish community (biomanipulation) has often been used to clear the water of algae and restore the aquatic plants in northerly locations, but it is again not clear whether this is equally appropriate at lower latitudes. 3. Eleven parallel experiments (collectively the International Mesocosm Experiment, IME) were carried out in six lakes in Finland, Sweden, England, the Netherlands and Spain in 1998 and 1999 to investigate the between-year and large-scale spatial variation in relationships between nutrient loading and zooplanktivorous fish on submerged plant and plankton communities in shallow lakes. 4. Comparability of experiments in different locations was achieved to a high degree. Cross-laboratory comparisons of chemical analyses revealed some systematic differences between laboratories. These are unlikely to lead to major misinterpretations. 5. Nutrient addition, overall, had its greatest effect on water chemistry then substantial effects on phytoplankton and zooplankton. Fish addition had its major effect on zooplankton and did not systematically change the water chemistry. There was no trend in the relative importance of fish effects with latitude, but nutrient addition affected more variables with decreasing latitude. 6. The relative importance of top-down and bottom-up influences on the plankton differed in different locations and between years at the same location. The outcome of the experiments in different years was more predictable with decreasing latitude and this was attributed to more variable weather at higher latitudes that created more variable starting conditions for the experiments.