This study assessed whether human food intake is regulated by negative feedback, directly or indirectly, from carbohydrate stores (glycogenostatic model). Six men were studied on three occasions during 7 d of whole-body indirect calorimetry, throughout which they had ad libitum access to one of three covertly manipulated diets: low fat (20% of energy as fat, 67% of energy as carbohydrate, and 13% of energy as protein; 4.80 kJ/g; LF), medium fat (40% of energy as fat, 47% of energy as carbohydrate, and 13% of energy as protein; 5.59 kJ/g; MF), or high fat (60% of energy as fat, 27% of energy as carbohydrate, and 13% of energy as protein; 7.04 kJ/g; HF). Energy intakes increased with percent fat (F[92,60] = 36.7; P < 0.001), producing average daily balances of -0.27, 0.77, and 2.58 MJ/d during the LF, MF, and HF diets, respectively. Changes in carbohydrate stores were attenuated by autoregulatory changes in carbohydrate oxidation. Carbohydrate balance showed a negative relation to the subsequent day's energy balance (t = 2.696; P = 0.0082) but explained only 5.5% of the variance. The relation for fat was positive (t = 5.245; P < 0.0001), accounting for 19.9% of the variance (stepwise regression). LF, lower-energy diets are more satiating than are HF-higher-energy diets, but carbohydrate stores per se did not entirely account for the change that diet composition had on energy intake. This study suggests that protein and carbohydrate have potential to reduce subsequent energy intake whereas there was no apparent reductive effect due to fat.