Aims. The aim of the study was to investigate factors hypothesized to influence the relapse process, with a focus on the role of self-efficacy, alcohol dependence and cognitive functioning. Design. The study was conducted in the context of a controlled trial of a relapse prevention programme. Subjects were assessed prior to treatment, at immediate conclusion of treatment and at 6- and 12-month follow-up. Setting. The study was conducted in an Alcohol Treatment Unit (ATU) in Scotland. Participants: Subjects were 60 male problem drinkers who were patients at the ATU. They were heavy drinkers, with corresponding high levels of alcohol dependence and alcohol-related harm. Measurements. The independent variables were posttreatment self-efficacy, alcohol dependence, cognitive functioning, level of depression and alcohol consumption prior to admission to treatment. The dependent variables were post-treatment drinking behaviour and functioning and time to lapse and relapse. Findings. Although the methodology does not allow identification of causality, support was found for the hypothesis that post-treatment self-efficacy was an intervening variable between treatment and outcome. Higher post-treatment self-efficacy predicted better outcome at 6-month follow-up and was associated with a reduced risk of lapse and relapse over the 12-month follow-up. Poorer cognitive functioning was significantly associated with being categorized as a problem drinker at 6-month follow-up and with higher risk of a lapse over the 12-month follow-up. Level of alcohol dependence did not predict outcome. Conclusions. It was concluded that post-treatment self-efficacy rating is a predictor of treatment outcome and time to lapse and relapse and that cognitive functioning is a predictor of treatment outcome and time to lapse.