This study aimed at predicting of live body weight from body measurements using stepwise regression analysis. Body
measurements data of 212 animals, Sohagi sheep flock (64 male and 148 female) were used. Body weight (BW) and four body
measurements were measured: heart girth (HG), height at withers (HW), height at rump (HR) and body length (BL). The
stepwise regression analysis was performed in order to retain the X variable(s) (the body measurements) that contribute
significantly (P < 0.05) to the variability in the dependent variable (BW). Results indicated that, there were high and positive
correlation coefficients between the body weight and all body measurements. The highest correlation coefficient (r=0.93) was
obtained between BW and HG and the lowest correlation coefficient (r= 0.88) was between BW and BL. All the studied body
measurements were entered into the model and through stepwise elimination procedure two of them were considered unfit in the
model (HR) and (BL). The two body measurements that best fit the model are heart girth (HG) and height at withers (HW),
accounting for 92% of the live weight in Sohagi sheep. Changes of R2 from the first model (R2=0.86, this model included HG
only) to the third model (R2=0.92), explained that, the most important variable in predicting BW is HG. The standardized
coefficient (Beta) is used to explain the contribution of each independent variable in the model. So, the most important variable
is HG (Beta = 0.92), this variable is the most important variable to explain the variability in BW. The prediction equation
explained that regression coefficient of BW/HG = 0.35, this means that when the heart girth increases by one unit (1cm), the live
body weight increases by 0.35 kg in sohagi sheep