BACKGROUND/OBJECTIVES: Predictive equations estimate post-bariatric surgery resting energy expenditure (REE), but lack accuracy assessment, especially for the remaining body mass. This study aimed to evaluate the agreement between indirect calorimetry and REE predictive methods. SUBJECTS/METHODS: It enrolled 226 females [median age 43.0 (36.2
50.4) years] who underwent mid- to long-term post-Roux-en-Y gastric bypass [median postoperative time 6.1 (4.0
9.0) years]. The measured REE (mREE) was obtained using indirect calorimetry, while the estimated REE (eREE) was derived from 18 predictive equations and an artificial neural network model. Analyses were performed for the total sample and body mass index (BMI) subgroups (<
30 kg/m² and ≥30 kg/m²). eREE within ±10% of mREE was considered accurate
Bland-Altman plots were performed to evaluate agreement. RESULTS: In the BMI <
30 kg/m² subgroup [n = 115
1372 ± 153 kcal (5744.3 ± 640.6 kJ)], mREE did not differ from four predictive equations
Henry [1371 ± 95 kcal (5740.1 ± 397.8 kJ), p = 0.922, bias -1.0 kcal (-4.2 kJ)] and Dietary Reference Intakes-Institute of Medicine [1382 ± 102 kcal (5786.2 ± 427.1 kJ), p = 0.315, bias 10.2 kcal (42.7 kJ)] equations showed better agreement and accurate prediction performance among BMI categories (79.1 and 82.6%, respectively). The BMI ≥ 30 kg/m² subgroup mREE [n = 111
1516 ± 186 kcal (6347.2 ± 778.7 kJ)] was significantly lower than all predictive methods and had higher bias and over-prediction, except for Mifflin-St Jeor equation [1523 ± 186 kcal (6376.5 ± 778.7 kJ), p = 0.469, bias 7.7 kcal (32.2 kJ)]. CONCLUSION: Equations for estimating REE show wide performance variation, with limited accurate options in this population, especially among those with BMI >
30 kg/m².