This study compared the Restraint Scale (RS) and the Cognitive Restraint Scale of the Three Factor Eating Questionnaire (TFEQ-CR) in their ability to predict negative-affect eating (disinhibition of restraint) in the laboratory. It was hypothesized that the RS would be a better predictor of disinhibited eating in the laboratory. Subjects (104 college women) were classified as either high or low on both scales, resulting in four separate groups. Subjects were then randomly assigned to either negative or neutral mood manipulations resulting in a 2 x 2 x 2 (RS x TFEQ-CR x Mood) design. A taste-test paradigm was utilized in which grams of crackers consumed following the mood manipulation was the dependent measure. No significant differences in laboratory food consumption were found between groups. Evidence was provided, however, suggesting that there are important differences between the two scales. The current study did offer some support for Lowe's (1993) three-factor model of dieting behavior, which differentiates between individuals who are currently dieting and those who have a history of frequent dieting and overeating.