American College of Sports Medicine (ACSM) metabolic equations are used to estimate energy expenditure (EE) of physical activity and prescribe aerobic exercise to meet EE requirements. Limited evidence supports their accuracy in sedentary adults with overweight or obesity during controlled exercise interventions. The purpose of this study was to compare EE estimated by the ACSM walking equation versus EE measured by indirect calorimetry during a 24-week aerobic exercise intervention, and identify potential modulators for their accuracy. Data from the exercising groups (8 or 20 kcal·kg body weight-1·week-1) of the E-MECHANIC study were utilized in this ancillary analysis (N = 103). Every 2 weeks for the initial 8 weeks and monthly thereafter, EE was measured via indirect calorimetry during absolute (2 mph, 0% grade) and relative (65-85% VO2peak) workload exercise. Resting metabolic rate, VO2peak, and body composition were assessed at baseline and follow-up. An EE offset factor (EOF) was calculated to express measured EE as a percentage of the estimated EE at each workload (EOF < 100% represents an overestimation of ACSM estimated EE). The accuracy of the equation decreased with increasing exercise workload (0.44%, 9.2%, and 20.3% overestimation at absolute, relative, and maximal workloads, respectively, at baseline) and overestimation of EE was greater after the exercise intervention. Furthermore, race, sex, age, fat mass, and VO2peak were identified as modulators for equation accuracy. Greater overestimation of EE was observed in Black compared to white females, particularly at lower exercise workloads. These findings support future efforts to improve the accuracy of metabolic equations, especially in diverse populations. Researchers should account for exercise efficiency adaptations when using metabolic equations to prescribe exercise precisely.