Over most of their natural northern Pacific Ocean range, pink salmon (Oncorhynchus gorbuscha) spawn in a habitat that was repeatedly and profoundly affected by Pleistocene glacial advances. A strictly two-year life cycle of pink salmon has resulted in two reproductively isolated broodlines, which spawn in alternating years and evolved as temporal replicates of the same species. To study the influence of historical events on phylogeographical and population genetic structure of the two broodlines, we first reconstructed a fine-scale mtDNA haplotype genealogy from a sample of 80 individuals and then determined the geographical distribution of the major genealogical assemblages for 718 individuals sampled from nine Alaskan and eastern Asian even- and nine odd-year pink salmon populations. Analysis of restriction site states in seven polymerase chain reaction (PCR)-amplified mtDNA regions (comprising 97% of the mitochondrial genome) using 13 endonucleases resolved 38 haplotypes, which clustered into five genealogical lineages that differed from 0.065 to 0.225% in net sequence divergence. The lineage sorting between broodlines was incomplete, which suggests a recent common ancestry. Within each lineage, haplotypes exhibited star-like genealogies indicating recent population growth. The depth of the haplotype genealogy is shallow ( approximately 0.5% of nucleotide sequence divergence) and probably reflects repeated decreases in population size due to Pleistocene glacial advances. Nested clade analysis (NCA) of geographical distances showed that the geographical distribution observed for mitochondrial DNA (mtDNA) haplotypes resulted from alternating influences of historical range expansions and episodes of restricted dispersal. Analyses of molecular variance showed weak geographical structuring of mtDNA variation, except for the strong subdivision between Asian and Alaskan populations within the even-year broodline. The genetic similarities observed among and within geographical regions probably originated from postglacial recolonizations from common sources rather than extensive gene flow. The phylogeographical and population genetic structures differ substantally between broodlines. This can be explained by stochastic lineage sorting in glacial refugia and perhaps different recolonization routes in even- and odd-year broodlines.