A previous study with amplified fragment length polymorphism (AFLP) fingerprints found no evidence of genetic impoverishment in the endangered Centaurea borjae and recommended that four management units (MUs) should be designated. Nevertheless, the high ploidy (6x) of this narrow endemic plant suggested that these conclusions should be validated by independent evidence derived from non-nuclear markers. Here, the variable trnT-F region of the plastid genome was sequenced to obtain this new evidence and to provide an historical background for the current genetic structure. Plastid sequences revealed little genetic variation; calling into question the previous conclusion that C. borjae does not undergo genetic impoverishment. By contrast, the conclusion that gene flow must be low was reinforced by the strong genetic differentiation detected among populations using plastid sequences (global FST = 0.419). The spatial arrangement of haplotypes and diversity indicate that the populations currently located at the center of the species range are probable sites of long-persistence whereas the remaining sites may have derived from a latter colonization. From a conservation perspective, four populations contributed most to the allelic richness of the plastid genome of the species and should be given priority. Combined with previous AFLP results, these new data recommended that five, instead of four, MUs should be established. Altogether, our study highlights the benefits of combining markers with different modes of inheritance to design accurate conservation guidelines and to obtain clues on the evolutionary processes behind the present-day genetic structures.