Research on predator impact in streams has generally ignored density—dependent effects. Although predator and prey behaviors are known to influence the impact of stream predators on local prey populations, no study has simultaneously addressed how predator impacts, and behavioral mechanisms contributing to those impacts, vary with prey density. I assessed the effect of a stream invertebrate predator (stonefly) on local prey (baetid mayfly) density across a gradient of six prey densities in in situ channels with natural substrata. Calculations of predator foraging rates and prey responses were based, in part, on an empirically based model assuming an exponential decline in prey densities. Total predator impact (measured as the negative natural log of the ratio of prey remaining in predator vs. predator—free channels) across prey densities followed a unimodal curve, with reduced predator effects at high and low, relative to intermediate, prey densities. Although direct predator consumption was more important than predator—induced emigration in explaining overall prey losses across all prey densities (65 vs. 35%), the relative importance of each loss process varied across individual prey densities. Total baetid drift increased with, and per capita drift was independent of, baetid benthic density in all predator treatments. Total and per capita drift rates of baetids were increased at least 3.5 times in the presence of predatory stoneflies. Few baetids were observed on the tops of rocks in either control or predator treatments during the day; however, baetid numbers on upper rock surfaces increased substantially at night and were significantly reduced by stonefly presence. The proportion of baetid populations on surface substrata was low and unaffected by predator presence during the day (0.3%) but was reduced in predator relative to predator—free channels at night (4 vs. 14%). Predatory stoneflies exhibited a type III functional response, and stonefly emigration decreased as initial baetid density increased. These results suggest that discrepancies in predator impacts between previous studies may be owing to prey density and reflect prey and predator behaviors that vary with prey density.