Here we employ multiple approaches (population genetics, past and present predictive modelling) to determine the population structure and the potential distribution in Portugal of the barbastelle, Barbastella barbastellus. Population structure was analysed by sequencing two mitochondrial DNA fragments: cytochrome-b and D-loop. The latter fragment showed almost no variation with only three haplotypes found. In contrast, 13 haplotypes were identified in cytochrome-b sequences although without a significant population structure despite the existence of unique haplotypes in the northern and southern regions. Predictive modelling showed that, under current conditions, the distribution of B. barbastellus is fragmented and highly dependent on the existence of native woodland, with the majority of the suitable areas located in the centre and north of the country, while the predicted occurrence in the south was very restricted. Our bioclimatic models for the Last Glacial Maximum also exhibited a similar pattern which, in agreement with the location of the unique haplotypes, suggests that some degree of isolation has existed between northern and southern populations. Nevertheless, the separation of these populations is not significant in the genetic analysis; hence, Portuguese populations should be managed as a single unit, where the maintenance of the existing mature native woodland is paramount for conservation. In conclusion, this study illustrates how combining genetic analysis with predictive modelling can help in the conservation management of a poorly documented species in a rapid and low-cost manner.