From 24 mating sets, 6300 fingerling of yellow perch (Perca flavescens) were stocked into one pond and equal numbers of progeny from six representative sets out of the 24 were stocked into each of two other ponds. After communal rearing for 21 months, total length and body weight were assessed for n=300 fish in each of the three ponds and molecular pedigrees were performed for each sampled individual to assign the progeny back to the original parents. The overall average number of alleles per locus was A=16.4 and observed and expected heterozygosities were Ho=0.88 and He=0.77 respectively. The mean weight of random samples and the top 10% fast-growing fish from the pond with all the sets was significantly greater than those from either of the two replicate ponds with six crosses. For the two replicate ponds, no significant differences were found in family rankings and assignment of the top 10% fast-growing fish, indicating that families with superior growth performance in one pond also exhibited the same superior growth performance in the replicate pond. However, there were no significant correlations detected in family mean weights of the top 10% fish between any two of the three ponds.