A vast body of research provides solid evidence of the effect that financial incentives have on the retirement decision of workers. Whereas these studies deliver valid results, they neglectthe fact that the retirement decision is most likely not homogeneous across all individuals. Studies conducted outside the field of economics showed that the type of task a worker performs affects the retirement decision, as well as his/her health condition. However, little is known about the potential relation between the performance of different tasks and the retirement age, and whether health mediates this relation.This study uses the data from the Dutch 2012 Liss Panel core study, in combination with data from the Value and effectiveness of work tasks study, which was conducted in 2013, to analyze the relationship between the performance of tasks, health and expected retirementbehavior. The analyses are in addition, performed separately for men and women. The findings of this study confirm that the retirement decision is not homogeneous across all individuals and that the retirement expectations differ with the tasks employeesperform in their job. In particular, the performance of tasks that require problem solving are significantly related to later retirement. This finding emphasizes the need for policy makers totake into account that not only financial incentives influence the retirement decision, but that this decision depends on factors, like the task composition of a worker. Furthermore, although the performance of manual tasks is related to bad health, and in turn bad health is related to early retirement, no mediation effect of health in the relationship between tasks and retirement is found. Finally, the results suggest that the relationship between job tasks and expected retirement age does slightly differ by gender.