The impact of predation by an avian predator (great crested grebe Podiceps cristatus) and a predatory fish (perch Perca fluviatilis) on a common resource, the small planktivorous fish smelt (Osmerus eperlanus) was studied in Lake IJsselmeer, The Netherlands, in July-October 1985-1988. At this time of the year the grebes are in simultaneous wing-moult and therefore temporarily flightless. Four factors limit the predation pressure by grebes: (1) an underwater visibility threshold of c. 40 cm Secchi depth is the lower limit at which large numbers attend the moulting site, (2) a lower density threshold of exploitable smelt biomass of c. 30 kg/ha determines the actual fishing areas, (3) a fish size threshold of 6.5 cm total length further limits prey availability, and (4) the vertical movements of the prey impose a strong (diel) time constraint on the birds, allowing only crepuscular foraging. Perch, the other main predator in this system, is less restricted in its foraging. Perch were found to be in direct competition with the grebes, preying heavily on the juvenile smelt. Over the 4 years of study 90% of all predation from 15 July to 15 October (72 233 kg/ha) was due to perch. The high level of predation by birds and fish was only possible due to a continuous immigration of smelt into the area (1.45 kg/ha per day). In three of the four years, however, the combined predation by fish and birds surpassed the immigration rate of the prey, which led to a strong reduction in smelt stocks in the study area. A conceptual model is developed to describe the different sets of constraints on the grebes' foraging. First, the state of eutrophication in relation to the weather condition determines the degree of algal blooms, and thus underwater visibility, in late summer. This is the major factor governing the numbers of grebes on the moulting area. Second, the size of the population of predatory fish determines the overall food availability (biomass and size distribution of smelt). These factors are partly interconnected and ...