OBJECTIVE To determine whether naltrexone is beneficial in the treatment of alcohol dependence in the absence of obligatory psychosocial intervention. METHODS Multicentre, randomised, double-blind, placebo-controlled trial. METHODS Hospital-based drug and alcohol clinics, 18 March 1998 - 22 October 1999. METHODS 107 patients (mean age, 45 years) fulfilling Diagnostic and statistical manual of mental disorders (4th edition) criteria for alcohol dependence. METHODS Patients with alcohol dependence were randomly allocated to naltrexone (50 mg/day) or placebo for 12 weeks. They were medically assessed, reviewed and advised by one physician, and encouraged to strive for abstinence and attend counselling and/or Alcoholics Anonymous, but this was not obligatory. METHODS Relapse rate; time to first relapse; side effects. RESULTS On an intention-to-treat basis, the Kaplan-Meier survival curve showed a clear advantage in relapse rates for naltrexone over placebo (log-rank test, chi(2)(1) = 4.15; P = 0.042). This treatment effect was most marked in the first 6 weeks of the trial. The median time to relapse was 90 days for naltrexone, compared with 42 days for placebo. In absolute numbers, 19 of 56 patients (33.9%) taking naltrexone relapsed, compared with 27 of 51 patients (52.9%) taking placebo (P = 0.047). Naltrexone was well tolerated. CONCLUSIONS Unlike previous studies, we have shown that naltrexone with adjunctive medical advice is effective in the treatment of alcohol dependence irrespective of whether it is accompanied by psychosocial interventions.