We examined biomass and abundance of organisms at the nonpiscivorous fish, benthic invertebrate predator and nonpredator, and microcommunity trophic levels at 30 stream sites in southern Ontario to test the food limitation, predation, and habitat limitation hypotheses of regulation of trophic structure. The presence of piscivorous fish at seven sites allowed us to contrast the predictions to the "bottom—up" (food limitation) and "top—down" (predator limitation) hypotheses directly. Also, we tested the size—efficiency hypothesis (size—selective predation) at the nonpiscivorous fish and benthic invertebrate trophic levels. In the presence of piscivores, the biomass of nonpiscivorous fish was lower and the biomass and abundance of benthic invertebrates were higher than in the absence of piscivores. Microcommunity biomass did not differ at sites with vs. without piscivores. Thus, we rejected the bottom—up hypothesis in favor of the top—down hypothesis. The influences of the top trophic level on lower levels decreased with "distance" from the top of the food chain. Also, benthic invertebrate predators were larger in the presence of piscivores, consistent with the size—efficiency hypothesis. The habitat limitation hypothesis was also supported at the benthic invertebrate and nonpiscivorous fish trophic levels. Invertebrate biomass and abundance were positively related to extent of riffles, whereas fish biomass and abundance were negatively related to extent of riffles. We speculate that this complementary distribution may be determined by fish predation on the invertebrates.