Objective Little is known about the effects of psychosocial work factors on objectively assessed mental health problems leading to medically certified absence. Only one study has evaluated the prospective effects of effort–reward imbalance (ERI) at work with regards to this outcome. The present study aimed to evaluate the effects of ERI on the incidence of medically certified absence for mental health problems.

Methods The study included 2086 white-collar workers (63.3% women) employed in public organisations in Quebec city. Participants were followed over a 9-year period. Medical absences from work were collected from employers’ files and psychosocial factors were measured using the ERI questionnaire. Cox regression models were used to estimate the incidence of certified sickness absence due to mental health problems that lasted 5 workdays or more, while controlling for confounders.

Results Workers exposed to ERI had a higher risk of a first spell of medically certified absence for mental health problems (HR=1.38, 95% CI 1.08 to 1.76) compared with unexposed workers. Low reward was significantly associated with a high risk among men (HR=2.80, 95% CI 1.34 to 5.89) but not in women. (HR=1.24, 95% CI 0.90 to 1.73). Effort at work had no effect on certified absence. All these effects were adjusted for potential confounders.

Conclusions ERI and low reward at work were prospectively associated with medically certified absence for mental health problems. These effects seem to differ by gender. Primary prevention that is aimed at reducing these stressors should be considered to help reduce the incidence of such severe mental health problems.