Current reformulations of the tension reduction hypothesis posit that only a subset of vulnerable individuals are at risk for drinking in response to negative affect. To further specify this model, this study examined the types of mood and social contexts under which affect and alcohol use are associated. Participants were 74 college students who completed repeated assessments of mood, alcohol use, friendship quality, and social support. A complex pattern of findings supported the moderating influences of gender, friendship factors, and the timing of behavior (i.e., weekends vs. weekdays) on the relation between affect and alcohol use. Young adults with less intimate and supportive friendships, as compared with their peers, showed risk for greater drinking following relative elevations in sadness and hostility. Such drinking episodes, in turn, predicted subsequent elevations in these same negative moods the following week. Gender differences in such a cyclical pattern of affect and alcohol use were found to vary across differing emotional experiences. Recommendations for a more refined theory of affect and alcohol use are discussed.