The aim of the present study was to investigate the sickness absence of insulin-treated diabetic employees. A case-control study using company sickness absence records for 1986 was performed on 59 diabetic patients and 59 control subjects matched by age, sex, and occupation. The patients had a similar frequency of absence to the control subjects (2.0 (range 0-7) and 1.7 (0-7) episodes yr-1, respectively). However, diabetic employees lost a larger number of working days (13.3 (0-101) vs 5.7 (0-52) days yr-1, p less than 0.03), and had a greater average length of episode of absence (5.6 (0-55) vs 2.5 (0-28) days, p less than 0.02). Males, manual workers, and those under 40 years of age accounted for significantly greater absence, but absence was also greater for diabetic employees than for control subjects in the other groups. Ten or more days were lost by 21 diabetic employees, but only 10 control subjects in 1986 (p less than 0.05).