A 7-year follow-up study is reported on the prevalent population of all insulin-treated diabetic patients (n=1499) as of 1 July 1973 in the Funen County, Denmark. The analysis of mortality was based on data from 395 dead and the remaining 1104 living patients. Males had a significantly higher mortality than females and a lower age at onset was associated with a significantly higher mortality. An analysis of the causes of death revealed a higher than expected number of deaths in all categories studied, although the excess mortality was highest for diabetes mellitus itself and cardiovascular diseases. Diabetes mellitus was not notified on 15% of the death certificates, and this under-reporting varied according to duration of the disease and place of death. It is concluded that studies based solely on death certificates will underestimate the mortality of diabetes mellitus, and that further longitudinal studies of well-defined, population-based patient groups are needed to evaluate the determinants of mortality in diabetes.