Abstract. By jointly considering patterns of genetic and life-history diversity in over 100 populations of Chinook salmon from California to British Columbia, we demonstrate the importance of two different mechanisms for lifehistory evolution. Mapping adult run timing (the life-history trait most commonly used to characterize salmon populations) onto a tree based on the genetic data shows that the same run-time phenotypes exist in many different genetic lineages. In a hierarchical gene diversity analysis, differences among major geographic and ecological provinces explained the majority (62%) of the overall G ST , whereas run-time differences explained only 10%. Collectively, these results indicate that run-timing diversity has developed independently by a process of parallel evolution in many different coastal areas. However, genetic differences between coastal populations with different run timing from the same basin are very modest (G ST Ïœ 0.02), indicating that evolutionary divergence of this trait linked to reproductive isolation has not led to parallel speciation, probably because of ongoing gene flow. A strikingly different pattern is seen in the interior Columbia River Basin, where run timing and other correlated life-history traits map cleanly onto two divergent genetic lineages (G ST Ïł 0.15), indicating that some patterns of life-history diversity have a much older origin. Indeed, genetic data indicate that in the interior Columbia Basin, the two divergent lineages behave essentially as separate biological species, showing little evidence of genetic contact in spite of the fact that they comigrate through large areas of the river and ocean and in some locations spawn in nearly adjacent areas.Key words. Allozymes, gene diversity analysis, life-history evolution, Pacific salmon, parallel speciation, run timing. The question of how rapidly and by what mechanisms adaptive differences arise among populations is of central interest to both evolutionary biologists and conservation biologists. Evidence is accumulating that evolution can occur at a rate high enough to be amenable to experimental observation within the lifetime of humans (Thompson 1998;Hendry and Kinnison 1999;Reznick and Ghalambor 2001). In addition, a number of recent studies have demonstrated the importance of parallel evolution, or repeated evolution of ecologically equivalent traits within a taxon (Reznick et al. 1996;Pigeon et al. 1997;Rundle et al. 2000;Johannesson 2001;Johnson 2001). Both types of studies raise questions about the importance of conserving existing life-history diversity and the likelihood that traits, once lost, will evolve once again-questions that are increasingly relevant to understanding the consequences of current rates of decline in biodiversity (Bernatchez 1995;Pimm and Raven 2000;Myers and Knoll 2001).Understanding the evolution of life-history diversity in salmon is particularly challenging, both because of the enormous complexity in life-history traits expressed by these species (Groot and Margolis 1991;Waples ...