Savanna woodlands and their associated species diversity and endemism are widely seen as declining through human impacts. Alternative views suggest that savanna ecosystems vary with a large number of biophysical factors, among which human impacts may be of relatively minor importance. This paper examines the debate with respect to Mkomazi Game Reserve in Northern Tanzania, where biodiversity has been inventoried and local resource use studied. It sets out the history of land use in Mkomazi and examines the available data on the area's plant, bird and invertebrate diversity. Comparative analysis is complicated by the paucity of data for other savannas in the same biogeographic zone, and by differences in sampling effort and methodology. Conservation literature and Tanzanian government documents present Mkomazi as one of the richest savannas in Africa, as a centre of endemism, and as threatened by deleterious impacts of human land use. Available data do not substantiate such statements. The paper examines implications of those perceptions for management, particularly eviction of resident pastoralists from the Reserve in 1988, and subsequent exclusion of reserve-adjacent dwellers. Conservation relies increasingly on reserve-adjacent people, and on prioritizing the allocation of scarce resources. There is an urgent need for rigorous studies of the implications of human land use in savannas, for better data on biodiversity, and for rigorous standards in the way those data are applied.