In the Southern Ocean, the at‐sea distributions of most predators of Antarctic krill are poorly known, primarily because tracking studies have only been undertaken on a restricted set of species, and then only at a limited number of sites. For chinstrap penguins, one of the most abundant krill predators breeding across the Antarctic Peninsula, we show that habitat models developed utilizing the distance from the colony and the bearing to the shelf‐edge, adjusting for the at‐sea density of Pygoscelis penguins from other colonies, can be used to predict, with a high level of confidence, the at‐sea distribution of chinstrap penguins from untracked colonies during the breeding season. Comparison of predicted penguin distributions with outputs from a high‐resolution oceanographic model shows that chinstrap penguins prefer nearshore habitats, over shallow bathymetry, with slow‐flowing waters, but that they sometimes also travel to areas beyond the edge of the continental shelf where the faster‐flowing waters of the Coastal Current or the fronts of the Antarctic Circumpolar Current occur. In the slow‐moving shelf waters, large penguin colonies may lead to krill depletion during incubation and chick‐rearing periods when penguins are acting as central place foragers. The habitats used by chinstrap penguins are also locations preferentially used by the commercial krill fishery, one of the last under‐developed marine capture fisheries anywhere on the planet. As it develops, this fishery has the potential to compete with chinstrap penguins and other natural krill predators. Scaling our habitat models by chinstrap penguin population data demonstrates where overlap with the fishery is likely to be most important. Our results suggest that a better understanding of krill retention and krill depletion in areas used by natural predators and by the krill fishery are needed, and that risk management strategies for the fishery should include assessment of how krill movement can satisfy the demands of both natural predators and the fishery across a range of spatial and temporal scales. Such information will help regional management authorities better understand how plausible ecosystem‐based management frameworks could be developed to ensure sustainable co‐existence of the fishery and competing natural predators.