Using a population-based cohort from 10 general practices in East Dorset, the mortality rate of diabetic patients compared to non-diabetic controls was investigated during 8 years follow-up. From a total population of 90660, 917 diabetic patients were identified; 693 (75%) with non-insulin-dependent (Type 2) diabetes and 224 (25%) with insulin-dependent (Type 1) diabetes. A control group of 917 non-diabetic subjects were selected, matched by age and sex. After 8 years, significantly more diabetic patients (334 or 36.4%) had died than controls (219 or 24%), (odds ratio (OR) 1.99, 95% CI 1.60-2.47). Compared with the controls, the odds ratio of all causes of mortality for diabetic men was 1.89 (CI 1.4-2.54) and for diabetic women 2.16 (CI 1.57-2.96). Compared with controls, the odds ratio for mortality from circulatory disease was significantly increased for diabetic patients 2.0 (CI 1.5-2.6) but mortality for respiratory disease or neoplasms was not significantly different (OR 0.7, CI 0.4-1.2 and OR 0.7, CI 0.6-1.0, respectively). Control data were lower than would be expected from national database data. The diabetic population had a significantly higher mortality than controls, both from all causes and circulatory diseases. Our data incidentally show the importance of appropriate controls for estimating the impact of a chronic disease.