Allozymes continue to be useful as genetic markers in a variety of studies; however, their utility often hinges on the selective neutrality of the allelic variation. Our study tested for neutrality between the two most common alleles (*100 and *110) at the cytosol nonspecific dipeptidase locus (PEPA-1*) in steelhead Oncorhynchus mykiss from Dworshak National Fish Hatchery in Idaho. We tested for differential growth and survival among fish with the * 100/100, *100/ 110, and *110/110 genotypes rearing in a hatchery or a natural stream. We repeated the study for two year-classes, using heterozygous (*100/110) adults to make the experimental crosses. This design avoided differences in family contribution among genotypes because each cross produced all three genotypes. We divided the progeny from each family into two groups. One group was reared in a hatchery for 1 year and then released for migration to the sea and subsequent return to the hatchery as adults. The other group was released into a natural stream and monitored for 3 years. We found no significant differences in size or survival among PEPA-1* genotypes for either the naturally reared fish or the hatchery-reared fish immediately prior to release as smolts. For females, survival to returning adult also was similar among genotypes; however, hatchery-reared males with the *110/110 genotype returned at a higher rate than did males with the *100/ 100 genotype; heterozygous males were intermediate. These results indicate that selection occurs at the PEPA-1* locus or at one or more loci tightly linked to it. The finding of nearly equal frequencies for these two alleles in the source population suggests that selection differentials among genotypes reverse or vary from year to year; otherwise, steady directional selection would drive the *100 allele to low frequencies or extinction. Locus PEPA-1* seems inappropriate for genetic marks in studies of steelhead that span the full life cycle and probably should be avoided for any portion of the life cycle. Inferences about gene flow and population structure from studies that are substantially influenced by PEPA-1* allele frequencies might be misleading. ?? Copyright by the American Fisheries Society 2006.