The distribution of many cephalopod, crustacean and fish species in the Southern Ocean, and adjacent waters, is poorly known, particularly during times of the year when research surveys are rare. Analysing the stomach samples of satellite-tracked higher predators has been advocated as a potential method by which such gaps in knowledge can be filled. We examined the viability of this approach through monitoring wandering albatrosses Diomedea exulans at their colony on Bird Island, South Georgia (54°S, 38°W) over the winters (May–July) of 1999 and 2000. At this time, these birds foraged in up to three different water-masses, the Antarctic zone (AZ), the sub-Antarctic zone (SAZ) and the sub-Tropical zone (STZ), which we defined by contemporaneous satellite images of sea surface temperature. A probabilistic model was applied to the tracking and diet data collected from 38 birds to construct a large-scale map of where various prey were captured. Robustness/sensitivity analyses were used to test model assumptions on the time spent foraging and relative catch efficiencies and to evaluate potential biases associated with the model. We were able to predict the distributions of a wide number of cephalopod, crustacean of fish species. We also discovered some of the limitations to using this type of data and proposed ways to rectify these problems.