A random sample of the middle-aged population of Malta was studied in a diabetic survey in 1981. Among the 1537 survey responders, in 659 men and 878 women aged 40 years or more the prevalence of diabetes was 16% in men and 18% in women, and that of IGT 5% in both sexes. By the end of 1985, 90 subjects (49 men: 41 women) had died. Mortality/100 (95% CI) was among subjects with normal, impaired glucose tolerance and diabetes in the age group 40-59 years 12 (5-19), 25 (23-73), and 61 (17-148), and in the age group of more than 60 years, 102 (68-136), 148 (59-237), and 178 (122-234), respectively. Age standardized mortality/1000 was in men 51, 28, and 100, and in women 34, 33, and 72 in the three categories of glucose tolerance, respectively. Among diabetic subjects aged 40-64 years the risk of death was increased seven-fold as compared with normoglycaemic subjects. Among men there was an inverse association between body mass index and mortality in all categories of glucose tolerance. Among women, no clear trend between body mass index and mortality was found. The relative risk of death for subjects with diabetes adjusting for age, sex, and body mass index was 2.2 (odds ratio: 95% CI 1.40 to 3.42) as compared with non-diabetic subjects. The age-adjusted survival curves for normoglycaemic subjects were similar for men and women. Among subjects with abnormal glucose tolerance the survival was slightly better in women than in men (p = 0.056). About half of the deaths were from cardiovascular disease and 7% from diabetes mellitus.