AbstractVideo image analysis (VIA), carcass shape and colour data were collected for 1211 lambs of known gender, breed type and carcass weight over a 1-week period using the VIAscan® system developed by the Australian Meat Research Corporation. Classification data (thickness of soft tissue over the 12th rib (GR measurement) and subjective conformation scores on a five-point scale of the leg, loin and shoulder) were assessed by an Agriculture and Agri-Food Canada grader after carcasses had chilled at 5°C for 3 to 6 h. Dissections into saleable meat yield (no. = 58) were performed after carcasses had chilled an additional 24 h. The timing of this study, which was dependent on availability of the VIA equipment, influenced the age and type of lambs available for analysis. The majority of lambs evaluated were wool-breed wethers, age > 10 months, of average GR (15·7 (s.d. 0·2) mm) and muscle conformation (3·0, s.d. 0·1). VIA improved the prediction of saleable meat yield (R2 = 0·71, residual s.d. = 14g/kg) compared with the current classification system (R2 = 0·52, residual s.d. = 18 g/kg). Although prediction ofGR measurement by VIA resulted in a large residual error (residual s.d. = 2·4 mm), the proportion of waste fat (perirenal and subcutaneous) and bone dissected from the carcass was accurately predicted (R2 = 0·62, residual s.d. = 11 g/kg). Proportions of leg (R2 = 0·71, residual s.d. = 7 g/kg) and shoulder (R2 = 0·62, residual s.d. = 9 g/kg) primals were also accurately predicted by VIA, although there were no significant predictors for the proportion of the loin (P > 0·15). VIA improved the prediction of saleable meat yield compared with the current classification system. However collection of additional data including some from extremely lean or well muscled animals would be required before VIA could be recommended to classify lamb carcasses