To identify client characteristics that predict attendance at treatment sessions and to investigate the effect of attendance on outcomes using data from the UK Alcohol Treatment Trial. Logistic regression was used to determine whether there were characteristics that could predict attendance and then continuation in treatment. Linear regression was used to explore the effects of treatment attendance on outcomes. There were significant positive relationships between treatment attendance and outcomes at Month 3. At Month 12, these relationships were only significant for dependence and alcohol problems for those randomized to motivational enhancement therapy (MET). There were significant differences between groups in attendance, with MET clients more likely to attend than clients allocated to social behaviour and network therapy (SBNT). MET clients were also more likely to attend all sessions (three sessions) compared with SBNT (eight sessions). MET clients with larger social networks and those with confidence in their ability not to drink excessively were more likely to attend. SBNT clients with greater motivation to change and those with more negative short-term alcohol outcome expectancies were more likely to attend. No significant predictors were found for retention in treatment for MET. For those receiving SBNT, fewer alcohol problems were associated with continuation in treatment. Attending more sessions was associated with better outcomes. An interpretation of these findings is that, to improve outcomes, methods should be developed and used to increase attendance rates. Different characteristics were identified that predicted attendance and continuation in treatment for MET and SBNT.