There is conflicting evidence on the role of size-selective mortality in the demography of populations of young sockeye salmon (Oncorhynchus nerka). A previous field study suggested that increases in mean fork length and otolith size at emergence observed in a cohort of salmon between fry and smolt stages were due to size-selective mortality, such that fry with small fork lengths underwent higher mortality than larger individuals. However, a subsequent study used a simulation to show that such increases could not have been achieved by size-selective mortality without levels of survivorship of the cohort far lower than those that had been observed in the field. To account for field observations, the simulation study proposed that individuals with high metabolic rates had better survivorship, a process that was termed "selection for growth potential." Here, we use a simulation approach to show that size-selective mortality may result in shifts in mean fork length and otolith size at emergence comparable with those observed in the field at total mortalities within the range of estimates of natural values. The contrasting outcomes of earlier simulation work and the present study are probably due to the assumption by the former that otolith and fish size in young salmon were weakly correlated and the use of an inappropriate model of size-selective mortality. We conclude that size-selective mortality can explain the results of previous field studies and that little empirical evidence exists to support the hypothesis of selection for growth potential in cohorts of young salmon.