This article aims to measure the causal effect of reported mental health in 2006 (anxiety disorders and episodes of depression) on job retention in 2010. We used data from the Health and professional career survey (HPC) for this purpose. We used Bivariate Probit modelling to take into account potential endogeneity inherent to mental health variables, firstly estimating the job retention of individuals according to their mental health, and secondly mental health, particularly due to significant events during childhood. In addition, we controlled our results for socio-economic characteristics, employment, general health, risky behaviour and career path. We show that for men, the fact of reporting that they were suffering from a mental illness in 2006 was associated with a decrease in job retention four years later. No significant relationship of this type was identified in women, once their general health was taken into account. Robustness tests carried out over the 2007-2010 period and across different age groups confirm these results.