In Victoria (a southern Australian state) in 1995, Narcotics Anonymous had a small but growing membership providing an opportunity to study the early experience of new self-help members. Ninety-one new members were interviewed and 62 (68%) were reinterviewed after 12 months. Three measures of self-help participation were examined: service role involvement, step work, and stable meeting attendance. Lower prior involvement in treatment services and greater participation in self-help predicted subsequent self-help participation. Higher levels of secondary school education predicted service role involvement and longer periods in stable meeting attendance. Higher self-help participation through the 12 months prior to follow-up was associated with lower levels of hazardous alcohol use and higher emotional support at reinterview. Multivariate regression analysis suggested stable self-help meeting attendance and step work continued to predict reductions in hazardous alcohol use and improvements in social support, after controlling for a range of alternative predictors.