Our preliminary studies of the efficacy of lithium carbonate therapy for alcoholism under double-blind, placebo-controlled conditions demonstrated that alcoholics who took their assigned medication (lithium or placebo) for the first 6 months after discharge from an inpatient rehabilitation program were more likely to abstain from any alcohol use for 18 months following discharge than were alcoholics who took their medication erratically or not at all. Attendance at Alcoholics Anonymous (AA) meetings was also associated with medication adherence. We applied a structural equation model to data on the relationships between medication adherence, AA meeting attendance and abstinent outcome to clarify whether medication adherence or AA meeting attendance better explains the positive-outcome "adherence effect" we observed. Both medication adherence and AA meeting attendance evidenced direct and independent influences on abstinent outcome: medication adherence showed a small direct influence, and AA meeting attendance showed a much larger, independent influence.