This experiment involved a commercial evaluation of carcasses of extensively reared crossbred lambs. These were sired by Suffolk rams from either a selection line or a control line of a Scottish Agricultural College (SAC) selection experiment, in which selection wasfor lean growth, or by Suffolk ‘reference sires’ from an industry co-operative breeding scheme (SSRS). The lambs were slaughtered at a target live weight of 42 kg between June and October 1992. In total, 421 lamb carcasses were included in the evaluation, 173 from six selection-line rams, 193 from six control-line rams and 55from three SSRS rams. Each of the carcasses was visually appraised for estimated subcutaneous fat proportion and for conformation of the shoulder, loin and leg, as well as being classified using conventional Meat and Livestock Commission (MLC) scales for fat and conformation. Animals were slaughtered at an average age of 139·5 (s.d. 25·6) days and achieved an average cold carcass weight of 20·04 (s.d. 0·96) kg. Carcasses had an average estimated subcutaneous fat proportion of 122·3 (s.d. 22·4) g/kg — equivalent to MLC fat class 3L to 3H. Overall conformation scores, on a 15-point scale, averaged 8·63 (s.d. 1·80) points. Carcasses were cut into joints according to a leading supermarket specification. The weights of pairs of shoulder, flank, loin and leg joints were obtained for each carcass, as well as weights of bone and fat removed during jointing. Saleable meat weights and proportions averaged 15·31 (s.d. 0·76) kg and 765·9 (s.d. 10·0) g/kg respectively. At a constant carcass weight, the SAC selection-line progeny were significantly younger (-11 days), had a significantly higher carcass value (+£1·50), a significantly lower estimated subcutaneous fat proportion (-13 g/kg), and a significantly higher weight of saleable meat (+0·1 kg) and higher proportion ofsaleable meat (+4 glkg) than control-line progeny, but had lower conformation scores. SSRS progeny had similar growth and fatness to selection-line lambs, but had poorer conformation, and significantly more bone in the carcass than either of the SAC lines. When comparisons were made at a constant estimated subcutaneous fat proportion, all differences in conformation between SAC lines disappeared. However, SSRS progeny remained poorer in conformation. The SSRS rams werefrom afoundation generation of the scheme, and were not expected to be markedly superior for carcass characteristics. Carcass weight was byfar the most important predictor of weight of saleable meat, or leg and loin joint weights. Conformation and estimated fat proportion made only marginal improvements, if any, to the precision of prediction, with fat proportion being the more important of the two predictors.