Endemism and species richness are highly relevant to the global prioritization of conservation efforts in which oceanic islands have remained relatively neglected. When compared to mainland areas, oceanic islands in general are known for their high percentage of endemic species but only moderate levels of species richness, prompting the question of their relative conservation value. Here we quantify geographic patterns of endemism-scaled richness (''endemism richness'') of vascular plants across 90 terrestrial biogeographic regions, including islands, worldwide and evaluate their congruence with terrestrial vertebrates. Endemism richness of plants and vertebrates is strongly related, and values on islands exceed those of mainland regions by a factor of 9.5 and 8.1 for plants and vertebrates, respectively. Comparisons of different measures of past and future human impact and land cover change further reveal marked differences between mainland and island regions. While island and mainland regions suffered equally from past habitat loss, we find the human impact index, a measure of current threat, to be significantly higher on islands. Projected land-cover changes for the year 2100 indicate that land-use-driven changes on islands might strongly increase in the future. Given their conservation risks, smaller land areas, and high levels of endemism richness, islands may offer particularly high returns for species conservation efforts and therefore warrant a high priority in global biodiversity conservation in this century.biodiversity ͉ conservation ͉ human impact ͉ terrestrial vertebrates ͉ vascular plants W orldwide loss of biodiversity requires global conservation priority setting to channel limited international conservation resources to regions of highest conservation value and need for action (1-5). Approaches for using biological data as a component of priority setting vary but can largely be divided into (i) algorithmbased assessments such as minimum-area sets or gap analyses (2, 6-8) and (ii) index-based assessments using indices such as endemism or species richness as surrogates for the conservation value of a region (1, 4, 9). While many theoretical arguments underpin the strengths of algorithm-based assessments (2, 8, 10, 11), they require detailed distribution data that are only available for few taxonomic groups-almost exclusively terrestrial vertebrates (6, 9, 12-15), on which systematic conservation planning has thus relied increasingly in the past years. Such detailed data are not available for the vast majority of taxonomic groups on the global scale including vascular plants. Although great effort is being made in digitizing existing data from natural history collections for conservation purposes (16, 17), biodiversity loss is arguably proceeding more rapidly than the documentation of species distributions. Hence, an inventory-based approach, which forms a main basis for the present study, is a workable solution if global conservation planning is to be informed by vascular plants (18), a group of o...