This paper investigates whether on-the-job training has an effect on the employability of workers. Using data from the Netherlands we disentangle the true effect of training incidence from the spurious one determined by unobserved individual heterogeneity. We also take into account that there might be feedback from shocks in the employment status to future propensity of receiving firm-provided training. We find that firm-provided training significantly increases future employment prospects. This finding is robust to a number of robustness checks. It also holds for older workers, suggesting that firm-provided training may be an important instrument to retain older workers at work.