An experimental study was performed to disentangle parental and environmental effects on the growth of Atlantic cod Gadus morhua larvae and juveniles. Eggs were collected during the spawning season from spawning pairs (families) kept separately in specially designed spawning compartments. Newly hatched larvae were released simultaneously into two mesocosms of 2,500 and 4,400 m(3). Larval growth was monitored by sampling over a 10 week period, after which juveniles were transferred to on-growing tanks, where they were tagged and kept for up to 2 years. Maternal origin was determined by individual microsatellite genotyping of the larvae (n = 3949, 24 families) and juveniles (n = 600). The results showed significant positive correlations between egg size and larval size during the whole mesocosm period. Correlations, however, weakened with time and were no longer significant at the first tank-rearing sampling at an age of 9 months. Significant family-specific differences in growth were observed. The coefficient of variation (c.v.) was calculated in order to examine variation in standard length of larvae during the mesocosm period. Inter-family c.v. was on average 69% of intra-family c.v. Differences in zooplankton densities between the two mesocosms were reflected in larval growth, condition factor and c.v. Low food abundance appeared to reduce c.v. and favour growth of larvae that showed relatively slow growth at high food abundance. It is suggested that genetically determined variation in growth potential is maintained by environmental variability.