The caloric expenditure of 44 healthy, lean and obese women, 8 of whom were trained athletes, was measured by indirect calorimetry. Body composition was determined. Ages ranged from 18-65 yr and body weights from 43-143 kg. Stepwise, multiple-regression analysis was used to determine whether one or several variables best predicted the resting metabolic rate (RMR) of the women. The RMR and the thermic effect of food (TEF) were measured before and after the women consumed a mixed breakfast meal. The results showed that the currently available tables and regression equations overestimate the RMR of healthy women by 7-14%. Body weight was highly related to the RMR, and stepwise inclusions of various variables did not improve predictions of RMR. The slopes of the regression lines for nonathletes and athletes were significantly different. Regression equations for predicting RMR of women were developed: Nonathletes RMR = 795 + 7.18 kg WT; Athletes RMR = 50.4 + 21.1 kg WT. The range of RMR per kilogram body weight was wide for nonathletic, but narrow for athletic women. The metabolism of some lean and obese, nonathletic women was highly efficient, predisposing these women for developing and maintaining body fat. The TEFs were indistinguishable between nonathletic and athletic women, and formed a continuum from the lightest to the heaviest woman.