OBJECTIVE: The role of changes in Alcoholics Anonymous (AA) involvement and social networks in relation to abstinence following substance abuse treatment is studied. Specifically, the role of AA and network support for abstinence are examined in relation to their effect on changes in abstinence states between follow-ups. METHOD: Study sites were 10 representative public and private alcohol treatment programs in a northern California county. A recruitment of 367 men and 288 women seeking treatment were interviewed at intake and re-interviewed 1 and 3 years later to collect information about alcohol consumption, dependence symptoms, social support for reducing drinking, number of heavy drinkers in the social network and AA involvement. RESULTS: Significant predictors of 90-day abstinence at both the 1- and 3-year follow-up interviews included AA involvement in the last year, percentage of heavy or problem drinkers in the social network, percentage encouraging alcohol reduction and AA-based support for reducing drinking. Panel models estimated an increase in AA participation between 12 and 36 months posttreatment increased the odds of abstinence at 3 years by 35% above those at 12 months. The only significant mediator of AA’s effect on abstinence was the number of AA-based contacts supporting reduced drinking, which reduced the magnitude of the relationship by 16%. CONCLUSIONS: AA involvement and the type of support received from AA members were consistent contributors to abstinence 3 years following a treatment episode. The enduring effects observed from supportive networks demonstrate the importance of ongoing mechanisms of action that contribute to an abstinent lifestyle.