Access to further training among adults on the labor market is unequally distributed. In particular, workers in occupations that are likely to be replaced by machines in the future participate less in training. This is mainly because of the job tasks they conduct: workers conducting routine tasks are more likely both to be replaced and to receive less training. As a consequence, technological change may lead to further polarization on the labor market. However, this trend may be cushioned by educational and labor market institutions. In this article, to assess the impact of institutions, the association between job tasks and participation in non-formal job-related training is analyzed in 24 countries from the first and second rounds of the Program for the International Assessment of Adult Competencies (PIAAC). Multilevel regression analysis is applied to test the influence of macro variables on the task gradient in training. The results reveal that tasks are important predictors of training participation in all countries. Comparing the effects across countries, it is found that tracking in initial education increases inequality in training participation owing to abstract tasks. Vocational orientation, on the other hand, reduces the effect. Furthermore, collective bargaining coverage decreases the effects of tasks on training, whereas strong employment protection legislation increases them. This indicates that the inclusiveness of lifelong learning is already influenced by the initial educational system. Strong unions and dynamic labor markets further enhance access to additional training among vulnerable workers.