Multiple regression analyses showed that 1) each reason for drinking was predicted by a somewhat different set of demographic and psychosocial variables, and 2) the reasons for drinking, in turn, differentially predicted multiple indicators of quantity and frequency of alcohol consumption.