Assessing energy expenditure in obese people is problematic. Two questions arise: Can we predict energy expenditure accurately? Does actual or ideal body weight better predict energy expenditure? Two groups of obese subjects--65 hospitalized adults and 65 nonhospitalized adults--were studied. Both groups had actual body weights that were at least 30% above ideal body weights. For both groups, energy expenditure was measured by indirect calorimetry and calculated using the variables sex, actual and ideal body weight, age, and ventilatory status. All but three patients were receiving nutrition support by the enteral route (either orally or by tube) or by the parenteral route (with hypertonic dextrose, amino acid, and fat). The nonhospitalized subjects fasted during measurements of energy expenditure. Regression equations were derived to predict energy expenditure. Actual body weight better predicted energy expenditure than did ideal body weight. We conclude that actual weight should be used to predict energy expenditure in obese individuals.