OBJECTIVE: To determine effectiveness of advice from general practitioners to heavy drinkers to reduce their excessive alcohol consumption (35 U or more a week for men, 21 U or more for women). DESIGN: Randomised, controlled double blind trial over 12 months with interim assessment at six months. SETTING: Group practices (n = 47; list size averaging 10,000) recruited from Medical Research Council's general practice research framework, mostly in rural or small urban settings. PATIENTS: Patients recruited after questionnaire survey. Of total of 2571 (61.2%) of 4203 patients invited for interview who attended, 909 (35.4%) stated that in past seven days they had drunk above the limits set for study and had not received advice; they were randomised to control and treatment groups. INTERVENTIONS: Patients in treatment group were interviewed by general practitioner (who had had a training session) and received advice and information about how to reduce consumption and also given a drinking diary. END POINT: Study aimed at detecting a reduction in proportion of men with excessive alcohol consumption of 30% in treatment group and 20% in control group (for women 40% and 20%, respectively) with a power of 90% at 5% level of significance. In addition, corroborative measures such as estimation of gamma-glutamyltransferase activity were included. MEASUREMENTS AND MAIN RESULTS: At one year a mean reduction in consumption of alcohol of 18.2 (SE 1.5) U/week had occurred in treated men compared with a reduction of 8.1 (1.6) U/week in controls (p less than 0.001). The proportion of men with excessive consumption at interview had dropped by 43.7% in the treatment group compared with 25.5% in controls (p less than 0.001). A mean reduction in weekly consumption of 11.5 (1.6) U occurred in treated women compared with 6.3 (2.0) U in controls (p less than 0.05), with proportionate reductions of excessive drinkers in treatment and control groups of 47.7% and 29.2% respectively. Reduction in consumption increased significantly with number of general practitioner interventions. At one year the mean value for gamma-glutamyltransferase activity had dropped significantly more in treated men (-2.4 (0.9)IU/l) than in controls (+1.1(1.0)IU/l; t = 2.7, p less than 0.01). Reduction in gamma-glutamyltransferase activity tended to increase with number of intervention sessions in men. Changes in gamma-glutamyltransferase activity in women and changes in other indicators in both sexes did not differ significantly between treatment and control groups. CONCLUSIONS: If the results of this study were applied to the United Kingdom intervention by general practitioners could each year reduce to moderate levels the alcohol consumption of some 250000 men and 67500 women who currently drink to excess. General practitioners and other members of the primary health care team should therefore be encouraged to include counselling about alcohol consumption in their preventive activities.