Sire selection efficiency depends on the knowledge of accurate genetic para
meters. In France, artificial insemination (AI) sires are selected accordin
g to their own performances and those of their progeny, which are both reco
rded in test stations. Genetic parameters among progeny traits were estimat
ed using multi-trait REML (restricted estimation of maximum likelihood) ana
lyses in Charolais and Limousin breeds. The expected decrease in genetic va
riability algebraically calculated among progeny traits due to the selectio
n of sires was not observed. This selection was not a strict truncation. He
ritabilities of traits measured on progeny are moderate for growth traits,
morphology and live fatness scores (from 0.14 to 0.38) and slightly higher
for dressing percentage and carcass fatness score (0.50 and 0.44, respectiv
ely). Genetic correlations among progeny traits depended on traits, selecti
on programme and breed. Carcass weight and morphology were highly genetical
ly linked to corresponding live traits (live weight and conformation, respe
ctively). They can, therefore, be easily improved through indirect selectio
n in contrast to carcass fatness which has only a small genetic correlation
with live traits. (C) Inra/Elsevier, Paris.