Objectives Very little literature is available on the effects of psychosocial work exposures on mortality. The aim of the STRESSJEM project was to explore the prospective associations between these exposures and all-cause and cause-specific mortality.

Methods The STRESSJEM project was based on a French national representative sample of 798,547 male and 697,785 female employees for which data on job history on the 1976–2002 period were linked to mortality and causes of death data from the national registry. Job strain model exposures were imputed using a job-exposure matrix and three time-varying exposure measures were constructed: current, cumulative, and recency-weighted cumulative exposure. The prospective associations between these exposure measures and mortality were explored using Cox proportional hazards models.

Results 88,521 deaths occurred among men and 28,921 among women between 1976 and 2002. Low decision latitude, low social support, job strain, iso-strain, high strain, and passive job were found to be risk factors for all-cause mortality, cardiovascular mortality, suicide, and preventable mortality (including smoking- and alcohol-related mortality as well as external causes of death). The model with current exposure was the highest relative quality model. The fractions attributable to job strain were 5.64% and 4.13% for all-cause mortality, 5.64% and 6.44% for cardiovascular mortality, 5.29% and 9.13% for suicide, and 5.1% and 3.1% for preventable mortality, among men and women respectively (though non-significantly different from zero for cardiovascular and preventable mortality among women).

Conclusion Our findings underlined the role of the job strain model exposures on all-cause and cause-specific mortality. The burden of mortality attributable to these exposures may be substantial, especially for suicide among women. Prevention oriented towards the psychosocial work environment may reduce mortality among working populations.