Data were collected from the carcass dissection of 3,500 lambs from Border Leicester x Merino ewes which were mated to 133 rams from 5 Poll Dorset, 4 Merino, 2 Meridale, 1 White Suffolk and 1 Siromt studs. The lambs were slaughtered in 4 groups per year, over 4 years (1986 1989), with the slaughter group average liveweights being 30 kg and 35 kg for the ewes and 35 kg and 45 kg for the cryptorchids. The right side of the carcasses were divided into 6 primals (forequarter, flank, ribloin, loin, chump and hindleg) and the subcutaneous fat was removed from each primal and weighed. The hindleg, chump and loin were further dissected to separate the bone from the soft tissue. In 1988, 150 carcasses were selected for total carcass dissection. These carcasses had all 6 primals dissected into subcutaneous fat, bone and soft tissue. The weight of the lean meat content for both the total carcass and the hindquarter (chump, hindleg and loin) was estimated using 2 models. Model 1 used GR tissue depth and carcass weight to estimate lean meat weight and Model 2 incorporated factors (breed, sex, year, carcass weight and GR tissue depth) known to influence carcass composition. For estimating total carcass lean meat weight (TCY), Model 2 gave an improved estimate of lean meat (R2 = 0.984) compared to Model 1 (R2 = 0.982). This effect was too small to justify the use of Model 2 in abattoirs, rather than Model 1. Similar small improvements were observed when estimating hindquarter lean meat weight (HQY) (R2 = 0.970 and 0.973 for Models 1 and 2 respectively).