The present study investigated whether tension reduction expectancies were uniquely associated with self-reported mood following in-lab alcohol administration, given that little research has addressed this association. We also tested whether level of experience with alcohol, which may influence the learning of expectancies, moderated expectancy-mood associations. Regularly drinking college students (N = 145) recruited through advertisements completed self-report measures of positive alcohol expectancies, alcohol involvement, demographics, and pre- and post-drinking mood, and then consumed alcohol ad libitum up to four drinks in the laboratory. Regression analyses controlling for pre-consumption mood, blood alcohol concentration, and all other positive expectancies showed tension reduction expectancies to be a marginally significant positive predictor of negative mood post-drinking. This association was significant only for those who achieved lower blood alcohol concentrations in lab and those who reported less involvement with alcohol (i.e., lower typical quantity, heavy episodic drinking frequency, and years of regular drinking). Findings suggest that associations between expectations for mood and actual post-drinking mood outcomes may operate differently for less versus more involved drinkers. Clinical implications pertain to early intervention, when expectancies may be less ingrained and perhaps more readily modified.