Research has found that depressed moods increase eating among persons who are dieting and among those characterized by high levels of weight fluctuation. To determine whether eating improves depressed moods among persons who score high on the weight fluctuation factor on the Restraint Scale (Herman, et al, 1978), 72 college women consumed either a large amount, a small amount, as much as they desired, or no chocolate milk following application of the Velton Mood Induction Procedure. Subjects also completed the Multiple Affect Adjective Checklist (MAACL) before and after the induction procedure and after the experimental sessions. Two to four days later, all subjects again completed the Restraint Scale. An analysis of the results supported the psychosomatic hypothesis that the eating behavior of high weight fluctuation persons affects their depressed mood states while no such effect of eating on mood exists for low weight fluctuation subjects. High weight fluctuation subjects in the small consumption condition were significantly less depressed than those in the control condition. These findings suggest that eating may be reinforced by reductions in depessed mood only so long as the amount consumed remains small. (BL)