We describe the 5-year mortality and its risk factors in a cohort of elderly people with and without known diabetes mellitus. The PAQUID cohort was representative of the population older than 65 living in Gironde, south-west France. Potential mortality risk factors were collected during a baseline evaluation, using a health questionnaire, from 68.9% of a randomly selected sample of over-65s in 1988. A total of 237 subjects (8.5%) had diabetes. Annual review occurred for 5 years and cause of any death was ascertained from family doctors. After 5 years, 623 people (22.3%) had died, of whom 576 were non-demented; 30.0% of the diabetic group versus 20.3% of the non-diabetic group had died. Survival of the known diabetic group was lower than that of the non-diabetic group (p < 0.001), although this excess mortality was significant only in the 65 to 75 age range (relative risk 1.8; 95% confidence interval 1.2 to 2.8, p = 0.04). Cardiovascular mortality rate did not differ between the diabetic and non-diabetic groups (RR 1.2 [0.8-2.0]). Death related to neoplasia was significantly higher in the known diabetic group (RR 2.2 [1.2-3.3], p = 0.01). In the final model, integrating diabetes as a mortality risk factor in the total cohort, known diabetes at the baseline examination was an independent risk factor for mortality (RR 1.4 [1.0-1.8], p = 0.01), in addition to tobacco use, hypertension and functional dependency. These results confirm suggestions that diabetes increases mortality in the over-65 age group, perhaps with an adverse interaction with other pathology.