Shortly before discharge, 50 alcoholic inpatients completed an ‘objective’ measure of alcohol dependence, the Severity of Alcohol Dependence Questionnaire, and a ‘subjective’ measure, taken largely from questions previously used by Schaefer (1971). At six‐month follow‐up, there was a significant difference in the subjective measure between patients who had maintained harm‐free drinking and those who had returned to some form of harmful drinking, but no significant relationship was observed for the SADQ. It is concluded that relapse following treatment is better predicted by the subjective measure than by the objective measure of dependence and the implications of this for theories of rela