This study aimed at using a multivariate approach to describe the body structure of Djallonke sheep in northern Ghana and to determine which approach explains better the variation in body composition. Live weight (LW) and linear body measurements including heart girth (HG), neck girth (NG), chest depth (CD), height at withers (HW), rump height (RH), body length (BL) and pin-bone width (PBW) were obtained from 172 sheep aged between two and three years. The fixed effects of sex and age were tested using the general linear model (GLM) while the Nearest Neighbor method of Hierarchical Cluster Analysis was used to group body traits into clusters. Principal Component Factor Analysis was used to describe the variation in body traits where extracted factors were varimax rotated to enhance interpretability. The analysis of variance revealed significant (P < 0.01) differences in the morphological traits of the two sexes with higher values recorded for the male in all traits examined except in PBW, which was insignificant (P > 0.05). Age had no significant influence (P > 0.05) on the body traits. The sheep weighed 26.92 ± 0.89 kg averagely and had averages of other body measurements to be: 71.74 ± 1.23, 40.52 ± 0.79, 27.73 ± 0.52, 60.72 ± 0.86, 59.61 ± 0.87, 58.87 ± 1.06 and 12.81 ± 0.23 cm for HG, NG, CD, HW, RH, BL and PBW, respectively. The product moments of correlation were positive and significant (P < 0.05, 0.01; r = 0.18–0.99) for all pairs of traits. The body traits were categorized mainly into two clusters with the first cluster comprising the HG, HW, RH and BL while NG, CD and PBW formed the second cluster. The grouping of the traits was slightly different in Factor analysis where two underlying principal components (PC) were extracted to discern the variance structure of the Djallonke sheep. The first principal component which consisted of CD, HW, RH, BL and PBW explained 61.26 percent and the second, 12.92 percent thereby giving a maximum of 74.17 percent generalized variance in body measurements. The traits loaded on the first principal component are closely associated with bone growth hence describing the general body size conformation while the traits (HG and NG) on the second component seem to describe only the thoracic region. It can be concluded that both the Hierarchical Cluster analysis and the Factor analysis grouped body traits similarly but the later is to be recommended because of the additional ability of indicating the amount of variation explained by the developed factors.