Visual predation by nocturnal zooplanktivorous fish is assumed to be sufficiently limited to allow the evolution of vertical migration as means for predation avoidance by zooplankton. However, in situ measurements of predation rates by nocturnal fish are lacking. Most of our knowledge of this predation is based on stomach contents and laboratory experiments. Our objectives were to measure the in situ rate of zooplanktivory by a common, nocturnal coral-reef fish, Apogon annularis, and to assess, through laboratory and field experiments, the effects of light, prey density, and flow on the fish's predation. In situ, the fish selectively fed on large zooplankton (>1 mm) at a rate of 0.12 prey min-1, or 52% efficiency. Predation rates increased linearly with prey density, with no apparent effects of current speed and light intensity. Flume experiments indicated that feeding rates were saturating at a level corresponding to that found at 3 m depth on a moonless night and were negligible at >18 m. Predation rates by A. annularis were two orders of magnitude lower than those of diurnal fishes; however, the difference in carbon gain was largely offset by the greater size of nocturnal prey. Nocturnal vision in A. annularis is sufficiently sensitive to allow a remarkable detectability of large prey in its natural habitat. If this fish is representative of other nocturnal fishes, reconciliation between the potential for substantial nocturnal predation, as is demonstrated in the present study, and the evolution of vertical migration in large zooplankton could be related to the total abundance of those fishes and their depth distribution. Both parameters are as yet poorly known.