Individuals with severe mental disorders (SMD) have an increased risk of mortality from somatic diseases. This study examined whether this risk is different in persons with depressive disorders compared to those with other SMD (i.e. schizophrenia and bipolar disorder). In 1989, 20,625 employees of the French national gas and electricity company (15,011 men and 5614 women, aged 35-50) agreed to participate in the GAZEL cohort study. Three diagnosis groups were created based on sick leave spells from 1978 onwards: 1) no SMD, 2) depressive disorders and 3) other SMD. Dates and causes of death were available from January 1, 1990 to December 31, 2010. The association of diagnosis groups with mortality was estimated with hazard ratios (HR) and 95% confidence intervals (CI) computed using Cox regression. During a mean follow-up of 19.8 years, 1544 participants died, including 1343 from a natural cause, of which 258 died from cardiovascular diseases. After adjustment for age, gender, occupational status, alcohol consumption, smoking and body-mass index, participants with a history of sickness absence for SMD had a greater risk of natural mortality (HR: 1.24, CI: 1.08-1.43), cardiovascular mortality (HR: 1.49, CI: 1.08-2.05) and non-cardiovascular natural mortality (HR: 1.19, CI: 1.02-1.39). Compared to depressive disorders, other SMD were associated with an increased risk of natural mortality (HR: 1.94, CI: 1.17-3.22) and cardiovascular mortality (HR: 3.58, CI: 1.53-8.39). Job security and systematic medical follow-up may fall short of preventing premature death among workers with sickness absence due to SMD.