1. The carrying capacity of a site for migratory water birds, expressed in bird-days, can be of particular conservation value. Several attempts have been made to model this carrying capacity using ideal free distribution models such as, for instance, depletion models, in which the distribution is fully determined by exploitative competition. 2. In the tests of depletion models carried out so far, no alternative models were compared; rather, one specific model was tested. We tested whether bird-days were more in accordance with birds depleting the food resource (a1) until a critical food density which just enabled survival or (a2) until a threshold food density which renders the site as profitable as an alternative site; and birds (b1) satisfying their daily requirements or (b2) maximizing daily intake. 3. We studied Bewick's swans feeding on below-ground tubers of fennel pondweed in one part of an autumn staging site. In most years between 1995 and 2005, we measured tuber biomass densities around September, November and March, and counted swans daily during their stopover in October. 4. The best fit between observed and predicted bird-days was obtained by assuming that the swans were maximizing their daily intake and depleting the tubers until a threshold biomass density (which yielded the same energetic return as the alternative food source after accounting for a small part of the initial tuber biomass being out of reach of the swans). Also in line with daily intake maximization, the daily feeding time did not differ from 10 h day1, the value predicted for Bewick's swans b 5. Our results suggests that the applicable model to calculate carrying capacity may depend strongly on whether birds use a site to stopover or to winter, because it determines whether the birds are more likely to use a threshold or critical food density, and to behave as energy maximizers or satisficers.