• To obtain objective information about binge- and nonbinge-eating behavior, 12 women with bulimia and ten women without eating problems (controls) were asked to eat four meals in a structured laboratory setting, on separate nonconsecutive days. The same instructions were given to both groups. On two days, they were asked to eat a normal amount, and on two days, they were asked to eat as much as they could, ie, to binge. For each type of instruction, they were given a single- and a multiple-course meal. The patients ate significantly more than the controls when asked to binge, both on the multiple-course meals that they rated as typical of binges and on the single-course meals. When they were asked to eat normally, there was no significant difference in intake between patients and controls on either single- or multiple-course meals. After all meals, hunger ratings of patients were significantly higher than hunger ratings of controls. There was also a significant positive correlation between intakes of single- and multiple-course binge meals and an inverse correlation between intake of multiple-course binge meals in bulimic patients and their rating of how well they controlled their eating. Thus, a structured laboratory eating situation can be used to reveal differences between bulimic and normal individuals and has the potential for assessing clinical status and exploring mechanisms responsible for binge eating.