Linear regression equations with tritium or urea space and live mass as independent variables and muscle, bone and dissectable fat in bull carcasses were calculated. The accuracy of these equations was compared to othe15 similar to these, but where the independent variables were carcass mass and the specific corresponding tissue in the prime rib cut. The experimental procedures are described in the first article (Meissner et al, 1980a). The addition of live mass as second variable usually improved the accuracy of prediction from tritium or urea space measurement, even to the extent of 30 to 50% in the case of dissectable fat. However, this was still inferior to the accuracy realized from prime rib cut analysis, the latter which according to available literature has already proved to be of insufficient accuracy to allow for reliable quantification of within and between-breed differences in fat distribution. The most accurate predictions of muscle, bone and fat (subcutaneous, intermuscular, total and kidney fat) in this study, using water space measurement, showed a coefficient of variation of 9, 11, 56, 33, 35 and 42%for muscle, bone and the four fat depots mentioned respectively, when a live animal of 400 kg is considered. In particular, subcutaneous fat which is highly correlated to carcass grading, would be predicted unreliably (CV = 56%).