The white shark (Carcharodon carcharias) (Linnaeus, 1758), an iconic apex predator occurring in all oceans,1,2 is classified as Vulnerable globally3-with global abundance having dropped to 63% of 1970s estimates,4-and as Critically Endangered in Europe.5 Identification of evolutionary significant units and their management are crucial for conservation,6 especially as the white shark is facing various but often region-specific anthropogenic threats.7,8,9,10,11 Assessing connectivity in a cosmopolitan marine species requires worldwide sampling and high-resolution genetic markers.12 Both are lacking for the white shark, with studies to date typified by numerous but geographically limited sampling, and analyses relying largely on relatively small numbers of nuclear microsatellites,13,14,15,16,17,18,19 which can be plagued by various genotyping artefacts and thus require cautious interpretation.20 Sequencing and computational advances are finally allowing genomes21,22,23 to be leveraged into population studies,24,25,26,27 with datasets comprising thousands of single-nucleotide polymorphisms (SNPs). Here, combining target gene capture (TGC)28 sequencing (89 individuals, 4,000 SNPs) and whole-genome re-sequencing (17 individuals, 391,000 SNPs) with worldwide sampling across most of the distributional range, we identify three genetically distinct allopatric lineages (North Atlantic, Indo-Pacific, and North Pacific). These diverged 100,000-200,000 years ago during the Penultimate Glaciation, when low sea levels, different ocean currents, and water temperatures produced significant biogeographic barriers. Our results show that without high-resolution genomic analyses of samples representative of a species' range,12 the true extent of diversity, presence of past and contemporary barriers to gene flow, subsequent speciation, and local evolutionary events will remain enigmatic.