Body weight and eight linear body traits, namely heart girth, body and diagonal lengths, height at withers and at hip, width of loin and at pelvic bone, and depth at rear flank, were measured forthniylıtly on 32 Muturu bulls, 16 of which were born in the dry season (Muturu -D) and the remaining 16 in the rainy season (Muturu-R), and on 11 N'dama bulls. Correlations between all pairs of traits in all groups were high, positive and significant (r > 0.093). The first two principal components derived from the correlation matrix of the linear measurements, PCI and PC2 accounted for at least 98% of total variance in all cases and were regarded as, size' and 'shape' vectors, respectively. Whereas PC1 gave largest weight to heart girth, PC2 gave largest weight to two or more different other linear measurements in the three groups. N'dama had the best conformation, followed by Muturu -D and then Muturu - R. PC-based prediction models were more parsimonius than linear measure-based models and are considered preferable for selecting animals for "optimal" balance. Also, in addition to discriminating between the two different breeds, PC- based discriminant functions were able to discriminate among individuals within the same breed on the basis of putritional differences. These functions are therefore recommended for classifying animals according to different macro and micro criteria.