Substantial genetic variation is hypothesised to be necessary for the long-term survival of species. Therefore, a major aim in conservation is to maintain and restore variation in small and endangered populations. However, in most populations the amount of genetic variation and, thus, the potential threats posed by limited variation are unknown. In the present study, we assess genetic variation, both at 10 microsatellite loci and at the major histocompatibility complex (MHC), in three closely related Acrocephalus warbler species with contrasting demographies. We found that the recently bottlenecked, island endemic, Seychelles warbler (A. sechellensis; SW) has substantially reduced microsatellite and MHC diversity compared to the widespread great reed warbler (A. arundinaceus; GRW). In contrast, another endangered species with a small breeding range, the Basra reed warbler (A. griseldis; BRW), harboured as much variation as did the GRW. This suggests that significant genetic variation and, therefore, adaptive potential, remains in the BRW - a situation that should hold as long as its habitat and numbers are maintained. Our study is one of the first to assess genetic variation at both 'non-critical' microsatellite markers and 'critical' MHC loci within endangered species. The two types of loci provided a similar picture of the genetic variation in the species we studied, but this has not been the case in studies of some other species. Using a combination of specific functional loci and genome-wide random markers appears to be the best way to assess the threat posed by reduced genetic diversity.