OBJECTIVE To evaluate resting energy expenditure compared to predicted energy expenditure in patients with cervical or ovarian carcinoma who require specialized nutritional support. METHODS Women with biopsy-proven cervical or ovarian carcinoma referred to the Nutrition Support Service were studied. Resting energy expenditure was measured by indirect calorimetry and compared to predicted energy expenditure (PEE) as determined by the Harris-Benedict equation for females. RESULTS Sixty one patients were studied. Patients with ovarian cancer (n = 31) had a significantly higher measured resting energy expenditure (% PEE) than patients with cervical cancer (109 +/- 18% vs. 98 +/- 16%, p < 0.02, respectively). This difference in measured resting energy expenditure between groups could not be explained by differences in the extent of disease, nutritional status, body temperature, or nutrient intake between groups. A greater proportion of patients with ovarian cancer were hypermetabolic (> 110% of predicted) in comparison to patients with cervical cancer (55% vs. 13%, p < 0.01, respectively). Measured resting energy expenditure varied between 53% and 157% of predicted for the entire population. CONCLUSIONS Ovarian cancer patients are more hypermetabolic than cervical cancer patients. The Harris-Benedict equation for females is a unreliable estimate of caloric expenditure in patients with cervical or ovarian cancer receiving specialized nutritional support.