Male medical in-patients were screened by interview. Those revealing alcohol-related problems or excessive consumption reaching a criterion were allocated at random either to no intervention or to brief counselling and a booklet (n = 156). Follow-up (n = 134) by a blind interviewer after one year with blood tests revealed significantly greater improvement in problem scores and gamma GT (n = 124) in the intervention group than controls and a significantly greater number of cases who reached a criterion of definite improvement confirmed by blood test or relative's information. Patients were in hospital for conditions which may or may not have been alcohol-related. The screening was brief. The intervention was motivational and behavioural. The medical context was believed to be important.