This population-based survey aimed to determine the prevalence of known diabetes mellitus on 31 December 1986, and to assess all-cause mortality in the subsequent 5 years (1987-1991) in Verona, Italy. In the study of prevalence, 5996 patients were identified by three independent sources: family physicians, diabetes clinics, and drug prescriptions for diabetes. Mortality was assessed by matching all death certificates of Verona in 1987-1991 with the diabetic cohort. Overall diabetes prevalence was 2.61% (95% confidence interval 2.56-2.67). Prevalence of insulin-dependent and non-insulin-dependent diabetes mellitus was 0.069% (0.059-0.078) and 2.49% (2.43-2.54), respectively. Diabetes prevalence sharply increased after age 35 years up to age 75-79, and finally declined. Prevalence was higher in men up to age 69 years, in women after age 75 years. Of the diabetic cohort 1260 patients (592 men, 668 women) died by 31 December 1991, yielding an overall standardized mortality ratio of 1.46 (CI 1.38-1.54). Even though the differences narrowed with age, mortality rates in the diabetic cohort were higher than in the non-diabetic population at all ages. Women aged 65-74 years showed observed/expected ratio higher than men (2.27, CI 1.92-2.66, vs 1.50, CI 1.30-1.72), while in other age groups the sex-related differences were not significant. Pharmacological treatment of diabetes was associated with an excess mortality, while treatment with diet alone showed an apparent protective effect on mortality (observed/expected ratio 0.73, CI 0.58-0.92).(ABSTRACT TRUNCATED AT 250 WORDS)