Studies were carried out in three lakes of strongly contrasting trophic state to evaluate how the effects of zooplankton on phytoplankton vary as a function of lake productivity. Chlorophyll and total P concentrations differed by 2-3 orders of magnitude among ultra-oligotrophic Lake Tahoe, meso-oligotrophic Castle Lake, and strongly eutrophic Clear Lake. Three experimental designs involving short-term (4 d) manipulations of nutrients, ambient zooplankton, and the common crustacean grazer Duphnia were performed in each lake. Algal responses were assessed at both the "aggregate," community (Chl, primary productivity, nutrient status) and species levels.Experiments in Lake Tahoe revealed a high degree of nutrient sensitivity but negligible grazing impacts by the very low densities of ambient zooplankton. Daphnia grazing had substantial impacts on Tahoe phytoplankton, however, indicating high susceptibility to grazing in this assemblage of relatively small-sized algal species. Castle Lake experiments revealed strong direct and indirect impacts of both ambient zooplankton and Duphnia on the nutrient-limited algal assemblage. The cyanobacteria-dominated algal community of Clear Lake was resistant to grazing impacts, responding relatively weakly only to the highest densities of Duphnin. Good correspondence between estimates of algal biomass made from chlorophyll measurements and microscopic examination were obtained, and, in Castle Lake, concordance between estimated community productivity tumover times made from species-specific growth determinations and community productivity measurements was observed. The contrasting responses to experimental zooplankton manipulations in the three lakes provide support for the view that the coupling between zooplankton and phytoplankton is strongest in lakes of intermediate productivity and imply that food-web alterations at the top of the food web are most likely to propagate to the level of the phytoplankton (and therefore lake water quality) in lakes of moderate trophic status.The relationship between nutrient loading and lake trophic state or productivity (Vollenweider 1976) is one of the most wellfounded empirical properties of lakes. Having established this relationship and used it in lake management with success, limnologists increasingly have focused their attention on the considerable amount of unexplained variance which remains in these empirical relationships.One explanation proposes that the chain of predatory interactions within the pelagic food web can strongly alter phytoplankton abundance by ' Present address: