A physiological distinction between eating occasions may help account for contradictory findings on the role of eating frequency in energy homeostasis. We assessed this issue using a midafternoon eating occasion known in France as the goûter that often consists of snack foods. Among the 24 male subjects, 8 habitually consumed four meals per day, i.e., were usual goûter eaters (GE) and 16 habitually took 3 meals per day, i.e., usual non-goûter non-snack eaters (NGNSE). All subjects were time blinded from lunchtime and had to request subsequent meals. Blood was continuously withdrawn and collected with a change of tube every 10 min until dinner request. During the session, 8 of the non-goûter eaters (NGE) were offered a snack 210 min after lunch and were designated as non-goûter snack eaters (NGSE) if they ate. Results showed that the goûter was preceded by high hunger scores and a linear decline in plasma glucose (-9.0+/-3.0%, P<.05) and insulin concentrations (-22.9+/-6.0%, P<.05). These profiles were not observed before the snack. The dinner of GE was requested later and was smaller compared to NGNSE, whereas the snack altered neither time of request nor energy intake (EI) at dinner. Among blood variables, leptin at the onset of eating was the only factor that was predictive of both intermeal interval and EI. The glucose and insulin profiles indicate that snacks should not be considered as meals in studies on the role of eating frequency in energy homeostasis.