Most experiments on effects of anti-predator behaviours use small enclosures and study behaviour and development of prey in the constant and immediate presence of predators. While useful, such studies are difficult to relate to natural situations where prey and predators are unrestricted. We describe a system designed specifically to estimate the impact of fish on foraging of larval Chironomus tentans under field levels of encounters with fish. The system involves raceways designed to control movement of fish over patches of chironomids so the impact of realistic, short-term encounters between predators and prey could be measured. Fish passes significantly reduced foraging activity, but the reduction was small (median of 1.16 min·fish event^-1·larvae^-1, or 6.5% of the time normally spent active) and did not depend on larval hunger nor fish size or species. Other laboratory experiments using large wading pools indicated 15 cm was the best estimate of the distance over which larvae respond to fish. We estimated foraging time lost in the field as frequency of encounter with fish (estimated from video recordings of fish traffic in 2 lakes) multiplied by time lost per encounter (estimated from laboratory studies). The most realistic estimates suggest larvae in the lakes would reduce their foraging time by 4–16% as a result of interactions with fish. Results suggest traditional techniques might overestimate the effects of predators on behaviour of prey.