Conservation planning and practice rely heavily on abundance information generated at local scales for decision-making. Unfortunately most distributional data are only provided as presence–absence records at regional or national scales and cannot be used for making local conservation decisions. To date, two methods have been proposed for estimating local abundances and area of occupancy from coarse-scale presence–absence data. Kunin (1998) proposed the use of the fractal method, which assumes that species distributions are self-similar across scales. He & Gaston (2000a), in turn, proposed the use of the negative binomial distribution (NBD) for estimating abundances from presence–absence data collected at different scales. Previously only plant data were used to test these models. In the present study, they were tested using distributional presence–absence, as well as abundance data for a selection of large herbivores that have a restricted distribution in the Kruger National Park, South Africa. The results illustrate that the fractal method tends to overestimate areas of occupancy, while the NBD method underestimates the areas of occupancy of all species. This means that there were significant differences between the predictions provided by the two methods. Both these methods rely on the relationship between the area of occupancy and the size of the area over which the survey is conducted, but describe this relationship in a different way. These techniques could prove useful in the field of conservation biology and this study illustrates that these methods can be used on large mammal species and deliver results that are consistent with previous studies on plants.