Using census data on three species of pygoscelid penguins, we tested the hypothesis that competition for food during the chick-provisioning stage of reproduction limits the number of conspecific seabirds in a region. This prey-depletion hypothesis predicts that a negative correlation should exist between colony population size and the total number of breeding pairs from other colonies within parental foraging range. We also tested whether or not a negative correlation exists between colony size and the population size of, or distance to, the nearest neighboring colony. Suitable data were available for Adélie, Chinstrap, and Gentoo penguins (Pygoscelis adeliae, P. antarctica, P. papua) along the coasts of Victoria Land and the Antarctic Peninsula, where major portions of these species' world populations nest. Results indicated that colonies were highly clustered, with small colonies grouped around one or two large ones, in turn spaced widely. Depending on species, two different patterns of geographic structuring were observed. For the Adélie and Gentoo penguins, no significant negative correlation existed between colony size and the total number of pairs breeding within parental foraging range of the reference colony; however, a significant negative correlation occurred at 150 and 200 km, well beyond foraging range. We found no relationship between colony size and size of or distance to the nearest neighboring colony. In contrast, for the Chinstrap Penguin, a significant positive correlation existed between colony size and total breeding population within the foraging range (50 km) but, as with the other two species, the correlations became more negative at greater distances. Moreover, a significant positive correlation existed between colony size and size of, but not distance to, the nearest colony in this species. We confirmed the hypothesis previously put forward that prey depletion by parents feeding chicks cannot explain size structuring of seabird colonies where breeding-season food supply is superabundant, as in polar regions. However, we also showed that prey depletion is not a necessary condition for negative size structuring. We suggest that if prey depletion occurs (by exploitative or interference competition), any manifestation in terms of colony distribution is overridden near to the colony by aggregating factors that originally led penguins to be colonial and philopatric, for example, social facilitation or predator avoidance. We further propose that geographic structuring is better explained by factors affecting the metapopulation (all breeders and nonbreeders associated with the colony cluster), especially during the prebreeding period, than by factors affecting chick-provisioning parents alone.