Adaptive changes in metabolism result in decreased energy requirements in AN. A retrospective study of 21 hospitalized female AN patients demonstrated that indirect calorimetry (IC) measurement of resting energy expenditure (REE) was significantly lower than REE calculated by the Harris-Benedict equation (HBE). The HBE was adjusted by multiple-regression analysis to reflect the hypometabolic state of AN, and the adjusted equation was prospectively validated in 37 hospitalized female AN patients. Refeeding requires an understanding of both baseline requirements and metabolic changes that occur during nutritional rehabilitation. In our present study, we prospectively evaluated changes in fasting and postprandial REE in 50 hospitalized female patients meeting DSM-IV criteria for AN. Baseline IC measurements of fasting and postprandial REE were obtained within three days of admission, and every two weeks thereafter. Mean fasting REE increased significantly from 72 (+/-11.7) to 83.2 (+/-12.6) percent of predicted (p < 0.001) during the first two weeks of hospitalization. Likewise, postprandial REE also increased significantly from 17.5 (+/-18.2) to 27.9 (+/-15.9) percent above fasting REE during the same time period (p < 0.01). Significant increases in both REE and postprandial REE persisted in patients requiring longer hospitalizations. Despite the fact that prescribed energy intake and triiodothyronine (T3-RIA) levels increased during refeeding, there was no significant relationship between postprandial REE and energy intake or T3 levels after baseline. We conclude that energy metabolism in AN adapts to semistarvation by a reduction in fasting REE. With refeeding there is a reversal of this adaptive function, demonstrated by an increase in both fasting and postprandial energy expenditure. The increase in postprandial REE is not related to energy intake or thyroid function.