Habitual meal frequency was assessed as a possible function of components of energy expenditure (EE) in human subjects. Fifty-six subjects participated (four categories differing in body composition): ten older women (fat-free mass (FFM) 42·0 (sd 6·3) kg, aged 59 (sd 2) years, BMI 27·5 (sd 6·9) kg/m2), fifteen younger women (FFM 45·5 (sd 5·2) kg, aged 34 (sd 10) years, BMI 21·9 (sd 2·3) kg/m2), twelve older men (FFM 56·8 (sd 5·9) kg, aged 62 (sd 4) years, BMI 25·7 (sd 3·3) kg/m2) and nineteen younger men (FFM 63·9 (sd 7·5) kg, aged 23·1 (sd 3·9) years, BMI 22·9 (sd 1·8) kg/m2). Measurements consisted of habitual meal frequency by validated food-intake diaries, physical activity by tri-axial accelerometers and resting EE by a ventilated hood system. Habitual meal frequency was expressed as a function of resting EE (including resting EE as a function of FFM), and of activity-induced EE, using regression analysis. FFM differed according to gender and age categories (P < 0·01). Physical activity level was higher in the younger men than in the other categories (P < 0·05). No relationship of meal frequency with the variables assessed was observed in subjects with a low FFM (the women). In the subjects with a medium FFM (the older men), meal frequency was positively related to resting EE (r2 0·4, P < 0·05), but not to the residuals of resting EE as a function of FFM, and inversely related to activity-induced EE (r2 0·3, P < 0·05). Resting EE explained 40% of the variation in meal frequency; adding activity-induced EE increased this to 60%. In the subjects with a high FFM (the younger men), meal frequency was inversely related to resting EE (r2 0·8, P < 0·0001) and to the residuals of resting EE as a function of FFM (P = 0·03), and positively related to activity-induced EE (r2 0·6, P < 0·0001). Resting EE explained 85% of the variation in meal frequency; adding activity-induced EE increased this to 89%. Habitual meal frequency was a function of components of EE, namely resting EE and activity-induced EE, only in subjects with a medium to high FFM (men). FFM-related differences in these relationships suggest a role of physical activity.