Variation in growth performances was analysed in 16 half-sib diploid and triploid families (from different sires) of rainbow trout (Oncorhynchus mykiss) and brown trout (Salmo trutta). At 1 year old stage (300 g) in rainbow trout, and 1.5 year old stage (320 g) in brown trout (3 months after transfer to seawater), 20 to 25 fish per family were individually tagged. Individual growth performances were followed, respectively, in freshwater for rainbow trout and seawater for brown trout up to their usual commercial size of 900 g and 3 kg, respectively. In rainbow trout, a significant genetic variability was observed among families in initial body weight, length and condition factor (P < 0.05). Mean daily growth rates of body weight during the freshwater rearing ranged from 0.70% to 0.80% between families but this difference was not significant. No significant genetic variability in final body weight (900 g) and length was observed. This was associated with an increased variability within families in these parameters and with the expression of a different genetic variability also suggested by a change in ranking between families. In brown trout, significant (P < 0.01) genetic variations in body weight, length and condition factor among families were observed at all stages up to commercial size. After transfer to seawater, a change in ranking of families for body weight was observed. Our results suggest that growth performance studies should be performed at a body weight close to commercial size, for 'filet portion sized' rainbow trout and large sized brown trout. At all stages, body weight, length and condition factor were significantly (P < 0.01) higher in diploid rainbow: trout than in triploid. This was observed also in brown trout; diploids showed a significantly greater body weight than triploids, but no significant difference was observed in final length at 3 kg. In both species, significant interactions between ploidy and family factors were observed for either body weight or length or condition factor at the initial stage studied but not at slaughtering. This result suggests that selection of diploid male breeders for growth performance would not be effective for their triploid progeny at 'pan-size' stages but effective at larger size.