ABSTRACTUltrasonic measurements of fat thickness and m. longissimus depth 25 and 50 mm lateral to the mid line at the last rib and 100 mm caudally to the last rib were made, before slaughter, on 273 ram lambs using two probes: the Scanogram and Krautkramer. The lambs were from three strains: a meat-type sire strain and two fecund dam strains. They weighed on average 37·3 kg at scanning, (range 29·0 to 51·3 kg), well within the weight range over which Canada Al-grade lambs are marketed in Canada. Their carcasses contained on average 437 g trimmed boneless meat per kg (s.d. pooled within strain = 24·1 g/kg).The precision with which the ultrasonic measurements combined with live weight at scanning predicted trimmed boneless meat (weight and proportion) was examined. Fat thickness measurements had no predictive value (P > 0·05). The residual standard deviations for the prediction of trimmed boneless meat in the half carcass (weight and proportion) from weight at scanning were 0·29 kg and 21·5 kg, respectively; the addition of the m. longissimus depth measurement reduced the residual standard deviations by 0·02 kg and 1·1 g/kg, respectively. The weight of trimmed boneless meat was predicted with more precision than the proportion of trimmed boneless meat. Strain, location of measurement site or type of instrument did not change the precision of prediction.It is concluded that prediction of trimmed boneless meat in young ram lambs based on live animal ultrasonic measurements made with the Scanogram or Krautkramer lacks the level of precision necessary for practical application.