We evaluated the mortality risk among 306 male alcoholics living in Osaka, Japan, at the time of initial diagnosis between 1972 and 1983, with regard to the cause of death, length of time from diagnosis, and participation in an alcohol abstinence self-help group. By the closing date on 1 March 1992, 110 of the 306 alcoholics had died, yielding an observed-to-expected (O/E) ratio of 4.5 [95% confidence interval (CI) = 3.7-5.4]. The alcoholics had significantly elevated mortality risks from all malignant neoplasms (O/E = 2.1, 95%CI = 1.2-3.3), esophageal cancer (O/E = 8.4, 95%CI = 1.7-24.5), diseases of the circulatory system (O/E = 4.4, 95%CI = 3.0-6.2), liver cirrhosis (O/E = 15.9, 95%CI = 10.2-23.6), diseases of the genitourinary system (O/E = 6.3, 95%CI = 1.3-18.5), and external death (O/E = 10.3, 95%CI = 6.3-15.8). The mortality risk from all causes still remained significantly high beyond the tenth year following initial diagnosis (O/E = 2.6, 95%CI = 1.0-6.2). The mortality risks from liver cirrhosis and external death (such as suicide) were highest within the first year following diagnosis, and were still high beyond the tenth year. A significantly high mortality risk from diseases of the circulatory system was observed between the first and ninth years, and the mortality risk from all malignant neoplasms was significantly elevated beyond 10 years following diagnosis. Alcoholics who did not join a self-help group soon after the initial institutional treatment had different cause-specific and time-specific mortality risks from those who did join a self-help group. These findings show the importance of long-term clinical follow-up of male alcoholics, taking into consideration the cause-specific mortality.