The relationship between spontaneous energy consumption and menstrual cycle was evaluated in 23 subjects who participated in one of two independent studies. Ad libitum intakes of experimental diets were measured by food weighing and bomb calorimetry for 56 or 42 d. Comparisons were made between each woman's mean energy during the 10 d before and after the onset of menstruation. The significant decline (364 kJ, or 87 kcal) between these two 10-d intervals was smaller than but consistent with findings from previous studies of data from food journals. In a separate analysis with time-series techniques, two distinct periods of elevated intake were identified (during the midluteal and midfollicular phases) that were independent of illness and menstrual symptoms. This pattern of food intake is discussed with reference to normal hormonal fluctuations. These findings confirm that menstrual cycle is a potential confounding variable that should be controlled in research on human food intake.