Various studies suggest that sea ranching of anadromous salmonids can result in domestication (increased fitness in the hatchery program) and a loss of fitness for natural production; however, the mechanism has not been characterized adequately. We artificially spawned hatchery and wild steelhead Oncorhynchus mykiss from the Clearwater River, Idaho, reared the resulting genetically marked (at the PEPA allozyme locus) progeny (HxH, HxW from hatchery females and wild males, and WxW) in hatcheries, and tested for differences in survival, growth, early maturation, downstream migration, and adult returns. Rearing treatments were mixed (crosses reared together) and separate (crosses reared separately from each other) at the hatchery of origin for the hatchery population where smolts are produced in one year, and at a nearby hatchery employing lower rations, lower winter temperatures, and two years of rearing to more closely mimic the natural life history (natural smolt age = 2-4 years). The hatchery population had been artificially propagated for six generations at the onset of our study. We found little or no difference in survival in the hatchery but substantially higher rates of growth and subsequent downstream migration for HxH than for WxW fish. Faster growth for HxH fish resulted in greater size at release which contributed to their higher migration rate, but other as yet uncharacterized traits also affected migration since the migration difference between crosses was apparent even within size classes. Growth of WxW fish was slower in the mixed than in the separate treatment indicating that WxW fish were competitively inferior to HxH fish in the hatchery environment. Incidence of precocious males was higher for WxW than for HxH fish in the separate but not in the mixed treatment. Incidence of HxH precocious males was similar between treatments. Apparently, the presence of HxH fish suppressed high incidence of early maturation by WxW males. A direct effect beyond the suppression of WxW growth by HxH fish was involved because the effect persisted within size categories. In-hatchery survival and growth of WxW relative to HxH fish may have been better with two-year rearing than in the standard one-year program (differences were consistent but small and non-significant); however, performance remained substantially worse for progeny of wild fish. Greater downstream migration for HxH than for WxW fish was primarily due to greater residualization for WxW than for HxH fish near the smolt release site rather than to immediate differential mortality. By August the residuals had lost condition compared to their condition in the hatchery the previous March, a month before release, and the residuals produced almost no smolts the following spring. Adult return rate was higher for HxH than for WxW fish for one year-class, consistent with the difference in downstream migration; only three adults (all WxW) returned from the other. Intermediate performance by HxW fish on growth, early maturation, downstream migration, and adult returns corroborated the genetic basis of the stock differences. Natural selection after release from the hatchery favored fish that performed well in the hatchery (grew fast, didn’t mature early, and excelled in other as yet uncharacterized traits) and genetically changed (domesticated) the wild population to resemble the hatchery population.