The objectives of the present study were to examine the effects of (1) ingesting mandatory snacks v. no snacks and (2) the composition of isoenergetically-dense snacks high in protein, fat or carbohydrate, on food intake and energy intake (EI) in eight men with ad libitum access to a diet of fixed composition. Subjects were each studied four times in a 9 d protocol per treatment. On days 1–2, subjects were given a medium-fat maintenance diet estimated at 1·6 × resting metabolic rate (RMR). On days 3–9, subjects consumed three mandatory isoenergetic, isoenergetically dense (380 kJ/100 g) snacks at fixed time intervals (11.30, 15.30 and 19.30 hours). Total snack intake comprised 30 % of the subjects' estimated daily energy requirements. The treatments were high protein (HP), high carbohydrate (HC), high fat (HF) and no snack (NS). The order was randomized across subjects in a counterbalanced, Latin-square design. During the remainder of the day, subjects had ad libitum (meal size and frequency) access to a covertly manipulated medium-fat diet of fixed composition (fat: carbohydrate: protein, 40:47:13 by energy), energy density 550 kJ/100 g. All foods eaten were investigator-weighed before ingestion and left-overs were weighed after ingestion. Subjective hunger and satiety feelings were tracked hourly during waking hours using visual analogue scales. Ad libitum EI amounted to 13·9 MJ/d on the NS treatment compared with 11·7, 11·7 and 12·2 MJ/d on the HP, HC and HF diets respectively (F(3,21) 5·35; P = 0·007, sed 0·66). Total EI values were not significantly different at 14·6, 14·5, 15·0 and 14·2 MJ/d respectively. Snack composition did not differentially affect total daily food intake or EI. Average daily hunger was unaffected by the composition of the snacks. Only at 12.00 hours did subjects feel significantly more hungry during the NS condition, relative to the other dietary treatments (F(3,18) 4·42; P = 0·017). Body weight was unaffected by dietary treatment. In conclusion, snacking per se led to compensatory adjustments in feeding behaviour in lean men. Snack composition (with energy density controlled) did not affect the amount eaten of a diet of fixed composition. Results may differ in real life where subjects can alter both composition and amount of food they eat and energy density is not controlled.