We measure gender differences in the causal impact of the 2006 self-assessed mental health status (anxiety disorders and depressive episodes) on job retention in 2010. We use data from the French Health and Professional Career Path survey. To control for endogeneity biases, we use bivariate probit models to simultaneously explain employment status and mental health. Anxiety disorders reduce men's job retention capacity by up to 12 percentage points (pp). Depressive episodes affect both genders almost equally (around 11pp). More severe cases of both mental health conditions are relevant in determining the capacity of individuals to remain in employment.