Anthropogenic natural habitat transformation presents the single most important threat to global biodiversity. Land cover data, based on Landsat Thematic Mapper (TM) imagery, were used to derive land use information for the Gauteng, Mpumalanga and Northern provinces of South Africa. The assessment integrated land use data with species presence data (15 × 15 minute grid cell resolution) for butterflies, mammals, birds and endemic vascular plants. The objectives of the present study were: (1) to identify areas at a regional scale where there is a possible conflict between biodiversity conservation interests and current land uses; (2) to investigate the influence of incorporating a land use constraint (LUC) into a conservation area selection algorithm, while taking cognizance of the existing reserve system; (3) to investigate the circumstances of species recorded within these conflict areas. Many grid cells identified as species richness hotspots, rarity hotspots or as part of the complementary network selected by the unconstrained algorithm were in reality largely transformed or modified. These areas should thus be avoided when striving to identify a viable conservation network. Although the LUC algorithm selected more grid cells to represent all species, it succeeded in increasing the percentage of natural vegetation within the selected conservation network and highlighted areas where potential conflicts should be thoroughly investigated at a local scale.