This study shows that alcohol consumption enhances the prediction of candy consumption by implicit attitudes and at the same time decreases the predictive validity of cognitive restraint standards. Female participants were assigned to either an alcohol or a control condition and were then given an opportunity to taste candies. For participants in the alcohol condition, candy consumption was uniquely predicted by previously assessed implicit attitudes toward the candy. In contrast, candy consumption was primarily predicted by cognitive restraint (Three Factor Eating Questionnaire) in the control condition. Moreover, participants who consumed alcohol ate significantly more candy at the group level. These results indicate that alcohol increases the behavioral impact of impulsive determinants on eating behavior while disrupting the behavioral impact of reflective determinants. They further demonstrate that measures of implicit attitudes toward tempting stimuli add incremental validity for the prediction of self-control outcomes.