The effectiveness of cognitive-affective stress management training (SMT) as a drinking reduction program for heavy social drinking college students was investigated. The SMT package included muscle relaxation and meditation training, cognitive restructuring, and coping skill rehearsal during induced affect. Treated and control subjects rated the frequency and intensity of their anxiety, anger and depression and recorded their alcohol consumption on a daily basis over a 6-month period. SMT significantly reduced posttreatment daily anxiety ratings and was associated with changes in four of ten irrational beliefs and a shift toward more internal locus of control in treated subjects. Reduction in anxiety was no longer evident at the 2 1/2- and 5 1/2-month follow-ups. The men in the SMT group showed a significant decrease in daily drinking rates at posttreatment and at the 2 1/2-month follow-up, but drinking returned to baseline levels by 5 1/2 months for the group as a whole. However, significant improvement variance in daily moods and in drinking rates over all posttreatment periods was accounted for by individual difference variables in the trained subjects but not in the control group, suggesting that these cognitive, personality, and social support variables are associated with response to stress management training. Implications of these results for future prevention research are discussed.