Given current knowledge of mean stock-recruitment relationships and variation in past recruitment, yield of sockeye salmon (Oncorhynchus nerka) in Bristol Bay, Alaska, and Fraser River, British Columbia, might have been at least 100%-300% larger since 1950 than was actually achieved. Most of these gains would have been due to knowledge of optimum mean spawning stock size rather than specific recruitment anomalies; knowing all future recruitment anomalies at the time of each spawning stock choice would have likely only added 2%-5% to total catches. For some stocks, delayed density dependence (cyclic dominance) might have resulted in somewhat lower yields, but under optimal management would still have been higher than were achieved. Even given only estimates of optimum spawning stock size each year based on data available as of that year, but following fixed escapement harvest policy rules, managers could likely have achieved 30%-40% higher total yield. Key management experiments for the future will involve testing for cyclic dominance effects on two major stocks (Kvichak, Late Shuswap) to determine whether stocks with strong, delayed, density-dependent survival effects should be deliberately managed through fallow rotation strategies for juvenile nursery lakes.