Studies testing the tension reduction hypothesis of alcohol drinking have found contradictory results. Marlatt proposes instead that social drinkers drink to gain control over stressful situations. To test this, 60 male heavy social drinkers were threatened with a social evaluation, then some were given some control over this threat by hearing that they could critique the evaluation and evaluate their evaluator (retaliate). This group was expected to drink less than a threatened group not given control, or a nonthreatened group, and heavy drinkers were expected to be affected more than light drinkers. Then, to determine if people do retaliate after an unfavorable evaluation, subjects received bogus evaluations or no evaluation, and then evaluated "their evaluator." Those receiving favorable or unfavorable evaluations responded in kind, supporting the coping method provided in the evaluation-with-coping condition. However, although the group that was evaluated and given no control reported more anxiety and took fewer sips than the other groups, they did not differ in wine consumption. The results did not support either the tension reduction or the control hypothesis.