This paper addresses the question to what extent the strong positive correlation between education and training can be attributed to differences in individual-, job- and firm-specific characteristics. The novelty of this paper is to analyze previously unconsidered characteristics, in particular, job tasks and firm-fixed effects. The results show that once job tasks are controlled for, the difference in training participation between educational groups drops considerably. In contrast, firm-fixed effects only play a minor role. Moreover, we show that workers performing nonroutine tasks are considerably more likely to participate in training than workers with routine tasks.