We investigated camps of black Pteropus alecto and red flying-fox P. scapulatus in the early dry season of 1992 (April–June) in Kakadu National Park, a World Heritage Area in the wet-dry tropics of north Australia. Fifteen camp-sites were located and two types of camps were defined: main camps containing > 1000 animals and a significant component of young (P. alecto, n = 4; P. scapulatus, n = 2) and satellite camps of lesser size (P. alecto, n = 9; P. scapulatus, n = 3). The two species shared three camp-sites. All camps were in dense riparian vegetation (fresh and saltwater mangroves, paperbark forests, closed forest and bamboo), overhanging or adjacent to water inhabited by crocodiles, that would provide a protected microclimate and protection from predators. All camps were within 5 km or less of one or more unoccupied sites that were indistinguishable in landform or vegetation from camp-sites. In Kakadu NP riparian vegetation makes up about 6% of the park and is a non-limiting resource for flying-fox roosting. We propose that the initial colonization of camp-sites is essentially random when roost vegetation is non-limiting and that factors such as predation by humans, cyclones and fires determine the persistence and size of camps at the local level. At a broader temporal and spatial scale, interaction between the seasonal availability of forage and reproductive cycles is known to influence the location, size and structure of Pteropus camps. We advance the hypothesis that colonies of flying-foxes could be moved to new camp-sites, a strategy that would facilitate resolution of the frequently problematic management of these animals.