The heat production (HP) of broilers has been shown to increase through the years associated with higher breast meat yield, supporting the use of an energy system that better accounts for HP differences caused by changes in genetics, body composition (BC), nutrition, and environmental temperature. This study aimed to determine the influence of BC on fasting HP (FHP) and the sensitivity of models based on metabolic body weight (MBW) alone or including BC to determine the maintenance net energy of broilers. Four experiments (E1 to E4) were conducted to determine the influence of BC on FHP and develop prediction models, and two experiments (E5, E6) to validate them. In E1 to E4, a total of 2400 Cobb 500 broilers distributed into 96 floor pens, were fed three dietary treatments to induce BC differences. The FHP (indirect calorimetry) and body protein-to-fat ratio (PFR; Dual-Energy X-Ray Absorptiometry) were determined from 1 to 56 d. The metabolic allometric coefficient was estimated using non-linear modeling and multiple linear models (response variable: FHP) were adjusted to the data using JMP Pro 16.0.0. In E5, broilers fed one of six dietary treatments were tested for FHP and BC three times within 1-42 d. In E6, broilers of two genetic lines distributed into 54 pens (45 birds each) and fed three dietary treatments were also tested twice within 1-42 d. The allometric coefficient was determined to be not different from 0.75 (P >0.05). Four models developed (R2 >0.98), considering body weight, PFR, and age, were validated (R2 >0.92; error <3%). Results showed that the leaner the bird, the higher the FHP (P <0.0001). Models based on MBW alone underestimate (+10% dif.) the FHP of lean birds being the estimations unreliable (regression assumptions not met), but including BC allows detecting +18% FHP differences. FHP and body energy retention (ER), expressed per unit of feed intake, account for productive energy (Arkansas Net Energy; Ark NE).