Aims - To explore client characteristics that predict drinking outcomes using data from the UK Alcohol Treatment Trial (UKATT). Methods - Multiple linear regression was used to determine if there were any characteristics, measured before the start of treatment, that could predict drinking outcomes at three and 12 months, as measured by percent day abstinent (PDA) and drinks per drinking day (DDD) over the preceding 90 days. Results - Lower baseline DDD score and greater confidence to resist drinking predicted lower DDD at both three and twelve months following entry to treatment. In addition to baseline PDA and having greater confidence to resist heavy drinking, female gender, aiming for abstinence, more satisfaction with family life and a social network that included less support for drinking were predictors of percent days abstinent. Conclusions - Overall the strongest and most consistent predictors of outcome were confidence to avoid heavy drinking and social support for drinking. More predictors were identified for percent of days abstinent than for drinks per drinking day. For percent of days abstinent, a number of client characteristics at baseline consistently predicted outcome at both month three and month twelve.