The reliability of the doubly labeled water method for measuring total daily energy expenditure (TDEE) was measured in six adult women at a 6-mo interval. The contribution of random error to reliability was assessed by repeating each isotopic analysis. Physiologic variation was calculated from the difference between the total variance and the analytic variance. In addition, postabsorptive resting metabolic rate and thermic effect of a meal were measured to partition the physiologic variation in TDEE into its three components. No seasonal effect (summer-winter) in TDEE was detected in these six subjects. The within-subject coefficient of variation for TDEE was 7.8%, of which physiologic variation was 6.4%. Within-subject variation in physical activity was the major source of variation in TDEE. Review of the 16 published studies, in which at least two measurements of TDEE by doubly labeled water were performed in each subject, indicated that the reliability of the method is 7.8%, except under conditions of high water flux. Care, however, must be taken to ensure that the analytic contribution does not inflate this figure in any given laboratory.