Fifteen bulimic women (BN) and 19 healthy female controls (CO) were studied. The subjects were cross-over treated with either fluoxetine (FXT) or placebo during 4 days. They received, in randomized order, a breakfast containing pure carbohydrate (CHO) or a protein-rich (PROT) breakfast following day 3 and 4 of each treatment period. Twenty-nine different food items were offered for lunch. The fasting serum glucose and insulin concentrations and the fasting plasma tryptophan (Trp)/large neutral amino acid (LNAA) ratio were slightly higher in BN. The changes of these metabolic parameters in response to a CHO or PROT breakfast were similar in both groups. Across breakfast type, the plasma (Trp)/(LNAA) ratio at 120 min after breakfast was higher in BN. Total caloric intake at lunchtime was less in BN. In CO, less carbohydrate was selected at lunchtime following the CHO breakfast, an effect that was abolished by FXT. Breakfast type or FXT did not have any apparent effect on food intake at lunchtime in BN. This might indicate that bulimic subjects are less sensitive to serotoninergic stimuli than control subjects.