Theories of cascading trophic interactions provide specific predictions regarding the forces that regulate populations across trophic levels. Reducing predators in a food chain with three trophic levels should permit herbivores to increase, thus reducing primary producers. In a manipulative experiment involving a mulluscivorous fish (pumpkinseed sunfish, Lepomis gibbosus), freshwater snails, and periphytic algae, we tested this prediction. With 10 cages (3 ° 3 ° 2 m) in each of two lakes in northern Wisconsin, we generated three treatments: exclosures (fishless), enclosures (three pumpkinseed sunfish), and cageless controls, both in Mann Lake, which had high natural densities of pumpkinseed sunfish (HDP), and in Round Lake, which had low densities (LDP). During a 16—mo experiment, we quantified snail and periphyton biomass on plastic flagging within treatments during summer at 3— and 6—k intervals in 1986 and 1987, respectively. Predation by pumpkinseed sunfish reduced snail biomass on flagging, permitting periphyton biomass to increase, as compared to exclosures. As expected, periphyton biomass in cageless controls in Mann Lake (HDP) mirrored periphyton biomass in enclosures, whereas cageless controls in Round Lake (LDP) differed from exclosures. The periphyton assemblage changed dramatically with increasing grazing pressure. In enclosures (low grazing), filamentous algae and large, stalked diatoms dominated the periphyton assemblage, whereas in exclosures (high grazing), assemblages were dominated by small, adnate diatoms and a colonial, filamentous bluegreen alga (Gloeotrichia). In laboratory trials, snails preferentially fed on periphyton from enclosures (where grazing pressure had been low). Predation reduced snail density, but indirectly increased periphyton biomass, dramatically modifying species composition of the assemblage. Thus, in mesotrophic Wisconsin lakes, top—down interactions regulate the benthic, freshwater food chain.