We studied antipredatory responses of lotic mayfly (Baetis) nymphs in a factorial experiment with four levels of fish presence: (1) a freely foraging fish (the European minnow,Phoxinus phoxinus), (2) a constrained fish, (3) water from a fish stream, (4) water from a fishless stream. LargeBaetis nymphs drifted mainly during night-time in treatments involving either the chemical or actual presence of fish, whereas no diel periodicity was observed when the water was not conditioned with fish odour. The response was strongest when the fish was uncaged, which suggests that visual or hydrodynamic cues are needed in addition to chemical ones for an accurate assessment of predation risk. Fish presence had no effect on the drift rates of small nymphs. Instead, they increased their refuge use in the presence of a live fish. Chemical cues alone did not have any effect on the refuge use of any of theBaetis size classes. Our results indicate active drift entry by mayfly nymphs. Because predation pressure is spatially and temporally variable, nymphs must sample the environment in order to locate predator-free areas or areas with low predation risk. Drifting should be the most energy-saving way to do this. To avoid the risk from visually feeding fish, large individuals can sample safely (i.e. enter drift) only at night-time, while the small ones can also do this safely during the day. We suggest that, contrary to some earlier assumptions, mayfly drift is not a fixed prey response. Instead,Baetis nymphs are able to assess the prevailing predation pressure, and they adjust their foraging behaviour accordingly.