This study was conducted to develop equations to predict chemical composition of head, limbs, hide and blood in cattle. A database containing 335 animals from 10 trials, with 221 Nellore, 38 Nellore-Simmental and 76 Nellore-Angus (96 steers, 118 heifers and 121 bulls) animals was used. Models were constructed to estimate water, ether extract (EE), crude protein (CP), ash and macrominerals (calcium, phosphorus, magnesium, sodium and potassium) in the non-carcass parts of cattle. A stepwise procedure was conducted to determine the most significant variables within each model. Subsequently, a random coefficient model was used to construct the equation using studies as random effect, and sex and breed as fixed effects. The visceral fat was the most important variable in the prediction models developed, affecting EE and water in head and limbs; head CP; and hide and blood water. Carcass dressing affected head EE and water and water in the limbs. Carcass weight had significant influence on head CP and hide EE; it was also affected by hide percentage in empty body weight (EBW). The percentage of OV in EBW influenced hide water. Lastly, EBW had influence only on hide sodium. Sex affected the EE of head and limbs. No breed effect was observed on any of the equations obtained. The estimation of the composition of head, limbs, hind and blood is possible and recommended, once they do not have great relevance to the estimation of EBW composition.