In this study, the mortality of clinically treated Japanese alcoholics with diabetes mellitus was analysed. Fifty-one diabetic alcoholics without liver cirrhosis (DM), 23 diabetic and cirrhotic alcoholics (DM.LC), 44 cirrhotic alcoholics without diabetes (LC), and 354 alcoholics without either complication (AL) admitted to the National Institute on Alcoholism in 1985 were studied. Thirty-seven diabetics required insulin treatment, and 12 oral hypoglycemic agents. The 4.4-year survival and drinking status after discharge were studied in 1990. Stepwise logistic regression analysis showed that the estimated odds for death increased 8.10, 4.38, 3.70, and 3.27 times for the subjects with the alcohol misuse after discharge, DM, DM.LC, and LC, respectively. The 4.4-year survival rate of alcoholics who continued misusing alcohol was much lower in DM (26%,P < 0.0005) and LC (35%, P < 0.0001) than in AL (73%). The survival rate of those who stopped misusing alcohol was significantly higher in DM (90%, P < 0.0001), LC (88%, P < 0.0001) and AL (94%, P < 0.0005) than those who continued misusing alcohol. There was no significant difference in the survival rate between the alcoholics with DM.LC who continued misusing alcohol (50%) and those who stopped misusing alcohol (73%). In the dead patients, 56% of DM died unexpectedly or suddenly, whereas 71% of LD died of liver failure after hospitalization. These results suggest that diabetic alcoholics should be intensively educated for abstinence.