In this paper we aim to explain the distribution of red-breasted geese Branta ruficollis over different nesting habitats. To be safe from land predators red-breasted goose colonies were restricted to i) islands on rivers, ii) cliffs with peregrine falcons Falco peregrinus, and iii) the close proximity of snowy owl Nyctea scandiaca and roughlegged buzzard Buteo lagopus nests. Among years nest site availability varied by fluctuations in numbers of owls and buzzards in association with cycles in lemming abundance, but the total number of goose nests found in the study area did not vary. The distribution of geese, in combination with data on reproductive success, suggested a despotic mechanism: at cliffs, goose numbers were constant among years with an invariably high reproductive success, whereas large fluctuations in numbers on islands coincided with opposite trends in success. Apparently, geese nesting with owls or buzzards moved to the few islands present in the study area during years when these birds of prey were absent. Consequently, in such years the average density of geese on islands was more than twice as high as at cliff colonies (5.4 and 2.3 pairs per ha of foraging habitat, respectively). Colony size at cliffs may have been restricted by territorial behaviour of the geese, though there is evidence that, additionally, the host falcons also limited the number of nesting geese. Apparently rare in closely related species, we observed a negative density-dependent effect on reproductive success during the nest phase, and attribute this to limited food resources, reinforced by the high frequency of territorial interactions. This leads to the conclusion that, in addition to predation pressure, nesting density is an important agent in the link between lemming cycles and goose breeding success