We conducted a controlled, blinded, multicenter study of disulfiram treatment of alcoholism in 605 men randomly assigned to 250 mg of disulfiram (202 men); 1 mg of disulfiram (204 men), a control for the threat of the disulfiram-ethanol reaction; or no disulfiram (199 men), a control for the counseling that all received. Bimonthly treatment assessments were done for one year. Relative/friend interviews and blood and urine ethanol analyses were used to corroborate patients' reports. There were no significant differences among the groups in total abstinence, time to first drink, employment, or social stability. Among the patients who drank and had a complete set of assessment interviews, those in the 250-mg disulfiram group reported significantly fewer drinking days (49.0 +/- 8.4) than those in the 1-mg (75.4 +/- 11.9) or the no-disulfiram (86.5 +/- 13.6) groups. There was a significant relationship between adherence to drug regimen and complete abstinence in all groups. We conclude that disulfiram may help reduce drinking frequency after relapse, but does not enhance counseling in aiding alcoholic patients to sustain continuous abstinence or delay the resumption of drinking.