Grazing by two size classes of metazooplankton, mesozooplankton(Meso; >140 μm in size) and metazoan microplankton (Micro; 100–140 μn in size), was studied in a mesocosm experiment carried out off the SW coast of Finland, northern Baltic Sea, in late sununer. During the 3 week study, the mesocoam was manipulated periodically by the addition of nutrient (ammonium and phosphate) and fish predators (stickleback fry). During the experimental period, the mesocosm was sampled five times to measure metazooplankton grazing, using 5 μm pre-filtered and 14C-labelled natural nanoplankton as food. In spite of the presence of fish, Meso biomass increased throughout the experimental period. The biomass of Mean was composed mostly of different copepodite stages of Eurytemora affis and the cladoceran Bosmina longispina maritima, and that of Micro by biomass of copepod nauplii NIII-NVL. Owing to its larger biomass, Mean could exert a greater grazing pressure on the nanophytoplankton than could Micro. The biomass specific clearance rate (BSCR) was generally the same for both groups, occasionally higher in Micro. The BSCR increased during the first half of the study period, after which the clearance rates were depressed, coinciding with a sudden decrease in water temperature. The daily ingestion rate on nanophytoplankton (‰ of biomass as carbon) varied between 3 and 96‰ for Meso, and between 4 and 130‰ for Micro. When integrated over the study period, grazing on nanophytoplankton was estimated to provide about 60‰ of the carbon requirements of metazooplankton, respectively; thus, protists were probably important food. The overall gross growth efficiency for the whole metazooplankton community was estimated to be 32‰. Mean and Micro grazing was estimated to account for 8 and 2‰ of primary production, indicating that they were not able to control phytoplankton primary production enriched by nutrient additions.