Cascading effects of predators can affect ecosystem properties by changing plant biomass, distribution and assemblage composition. Using data from field surveys and whole-stream experiments we tested the hypothesis that predatory trout change assemblage composition of benthic algae in high-elevation streams mediated by grazer behavior. Field surveys revealed that the taxonomic composition of algal assemblages differed significantly between streams that contained trout and those that were fishless; but comparisons of palatable versus unpalatable algal taxa between fish and fishless streams were equivocal because of high natural variability. Therefore, we tested for a behavioral (non-consumptive) trophic cascade experimentally by adding brook trout chemical cues to six naturally fishless streams for 25 days and compared responses of grazers and algae to six reference streams without fish cues added. Algal response variables included rates of change in the abundance of three physiognomic categories, from most palatable (attached erect and prostrate diatoms) to least palatable (non-diatoms), as determined from food selectivity analyses of the most common grazers (mayflies and caddisflies). Fish cues did not affect the mean densities or changes in densities of total grazers or any individual grazer species. However, in streams where fish cues were added, rates of accrual of attached erect diatoms, which was the preferred algal type for the grazer most vulnerable to trout predation (Baetis), were higher and their densities increased significantly faster with increasing densities of this grazer species than in reference streams. Results of his experiment support the hypothesis that predator induced suppression of grazer foraging behavior, rather than cascading effects of top predators on grazer density, may contribute to variation in the composition of algal assemblages among streams by allowing proliferation of most palatable algal species.