Seventy-two mature, nonpregnant, nonlactating Hereford cows (400 kg) were utilized in a comparative slaughter trial to investigate the effects of carcass composition on the metabolizable energy (ME) required for maintenance in winter. Body condition score (CS), live weight (LW) and weight:height ratio (WTHT) were evaluated and compared as estimators of carcass composition in cows. Cows ranged in LW, CS and WTHT from 275 to 595 kg, 2.0 to 8.0 units and 2.29 to 4.62 kg/cm, respectively. Live weight, CS and WTHT predicted total carcass energy (TMCAL, r2 = .81, .85 and .83), carcass fat (FAT, r2 = .78, .82 and .80), carcass protein (PRO, r2 = .71, .74 and .70) and carcass water (WAT, r2 = .78, .71 and .77) with similar accuracy. When composition was expressed on a per unit weight basis, CS was superior to LW and WTHT as predictors of TMCAL/hot carcass weight, TMCAL/LW and FAT/hot carcass weight (r2 = .82, .60 and .64; .83, .58 and .62; and .82, .64 and .68, respectively). Forty-seven cows were individually fed a complete diet (2.50 Mcal ME/dry matter) in drylot for 114 d in yr 1 and 115 d in yr 2. Daily feed intakes were adjusted each week to maintain constant LW throughout the winter. Data were analyzed by fitting the model: ME intake = k-1 (carcass energy change) + f(CS)LW.75, where k = efficiency of ME use for carcass energy change and f(CS) = function of CS. Year as a class variable and the expression .1028 + .0234(CS) - .0025(CS)2 accounted for 41% of the variation in Mcal ME for maintenance/LW.75. Condition score was more closely related to carcass composition in mature cows than was LW or WTHT and cows in either a thin (CS = 3) condition or a fat (CS = 7) condition required less (4.4% and 8.9%) Mcal ME/LW.75 for maintenance than cows in moderate (CS = 5) condition.