Aims To examine the ability of the Impaired Control Scale (ICS) to predict outcome of moderation-oriented treatment for alcohol problems and to compare this predictive ability directly with that of a widely used measure of alcohol dependence, the Severity of Alcohol Dependence Questionnaire (SADQ). Design Prospective follow-up study. Setting Out-patient treatment centres. Participants A combined sample 154 problem drinkers taking part in two clinical trials of Moderation-oriented Cue Exposure in the UK and Australia. Clients were followed-up 6 (UK) and 8 (Australia) months after the end of reatment. Measurements Outcome was categorized by combining drinking behaviour at follow-up with changes on the Alcohol Problems Questionnaire from before treatment to follow-up. Controlling for research site, baseline scores on Part 2 of the ICS (substitution method) and the SADQ-C were entered in logistic regression analyses with three outcome dichotomies as dependent variables. Findings Five per cent of clients were abstinent at follow-up, 13% non-problem drinkers, 25% much improved, 24% somewhat improved and 34% unimproved. Location of treatment and ICS2 scores were significant predictors of whether or not clients achieved a successful outcome (abstinence or non-problem drinking). Using a cut-point of 25 on the ICS, two-thirds of outcomes were classified correctly as either treatment successes or failures. SADQ-C score was not a significant predictor of treatment outcome. Conclusions The ICS predicts outcome of moderation-oriented treatment among moderately dependent problem drinkers recruited mainly via newspaper advertisements. The ICS should replace the SADQ as the basis for advice to clients in this population of problem drinkers regarding whether or not a moderation goal of treatment should be pursued.