Self-efficacy for avoiding alcohol use predicts alcohol use after treatment. However, self-efficacy predicts outcome differentially depending on whether ratings are made before or after treatment. In order to increase the predictive validity of self-efficacy judgments, the hypothesized sources of self-efficacy were examined in the current study utilizing a college student population. Self-efficacy ratings for avoiding heavy drinking before and after self-monitoring of drinking behavior were examined in order to understand whether careful examination of current behavior would result in more informed self-efficacy judgments. Participants (N = 135) completed questionnaires that assessed self-efficacy, drinking behaviors, alcohol expectancies, and perceived normative alcohol use. Participants were assigned to either the control or self-monitoring condition. Self-monitors recorded drinking behaviors during the four weeks after the baseline assessment. All participants returned for a follow-up assessment four weeks later. Consistent with predictions based on social cognitive theory, heavy drinking, positive alcohol expectancies, and perceived norms of use were inversely related to self-efficacy. Hierarchical multiple regression analyses indicated that self-monitoring did not lead to stronger relationships between source variables and self-efficacy. Individuals who appeared to be overconfident in their self-efficacy judgments at baseline did not make more accurate ratings as a result of self-monitoring. Results from this study highlight potential sources of information individuals use in making self-efficacy judgments.