From 1983 all incident cases of diabetes in the age group 15-34 years in Sweden have been recorded prospectively. The aim of the present study was to assess the mortality pattern of cases reported for 1983-87 and followed until the end of 1987. Eighteen deaths were identified by linkage to the national death register. When comparing the mortality in the cohort with Swedish males and females in general, an excess mortality was found in all the groups studied. It is, however, less pronounced if the analysis is restricted to those with Type 1 diabetes (standardized mortality ratio (SMR) and 95% confidence interval = 2.1; 0.8, 4.6), Type 2 diabetes (SMR = 4.8; 1.3, 12.3) or Type 1 + Type 2 (SMR = 2.7; 1.3, 5.0). Eight (44%) of the deaths were in patients with secondary diabetes, a diagnosis that applied to less than 3% of the cohort. Alcohol abuse was prevalent in six cases and suspected in another two. Hypoglycaemia was established as a cause of death in only one case but could not be excluded in a further six. Only one death was associated with ketoacidosis. No valid support for an increased risk of the 'dead in bed' syndrome was found. We suggest that diabetes was decisively important for the death in two cases and less important in 10. In the remaining six cases the existing documentation precludes a proper judgement.