Abstract The sei whale (Balaenoptera borealis) presents a disjointed geographic distribution. Significant genetic divergence between sei whales from the North Atlantic (NA) and North Pacific (NP) have been recorded; however, limited samples from the Southern Hemisphere (SH) have precluded a global analysis. The largest recorded mass mortality event occurred in Golfo de Penas, southern Chile, providing a primary source of SH samples. A global mitochondrial DNA phylogeographic study was undertaken including 111 SH samples (98 from Golfo de Penas), and new sequences from the NP. A marked phylogeographic structure was observed between the three ocean basins' populations, with the highest ΦST values between NA–NP and NA–SH, and a lower but still significant value between SH–NP. This structure, and the haplotype distribution pattern, suggest a more recent separation between SH–NP and an absence of recurrent gene flow among all three populations. Only one potential migratory event was detected from SH to NA. These results support an ongoing phylogeographic differentiation between all populations, highlighting the identity of each and the importance of independent management. Therefore, each population should represent a separate management unit, which is important in the development of conservation strategies. Demographic analyses detected a recent reduction of population size in the SH and NA, but not in the NP. For all populations, estimated effective population sizes were higher than estimated census sizes. This suggests that most of the genetic diversity of pre‐whaling populations is still retained; however, continued reduction in population sizes could lead to future loss of genetic diversity. This study highlights the necessity of local protection and systematic monitoring of the Golfo de Penas sei whale population. In such a remote area the use of remote sensing techniques could be a cost‐effective option for evaluating the threat, periodicity and magnitude of mortalities and their effect on the local whale population. The integration of genetic and demographic approaches improves the understanding of the dynamics of populations, which is particularly useful in endangered species.