"The nature of work has changed remarkably in the last decades. This implies new risks and challenges for people at work. Although many studies have investigated the correlation between work and health, the causal effect of work on health is not yet well established. This project aims to identify the causal effect of work on health by exploiting information from several natural experiments in a regression discontinuity design. Furthermore, the project does not only focus on the decision to work or not; other work dimensions that may affect health, such as the number of working hours and the quality of the job, are also considered.This project goes beyond existing studies in three aspects. First, the empirical strategy is based on several different policy changes, which generate exogenous variation in employment rates, the number of hours worked, and job quality. Using these natural experiments I can identify an unbiased estimate for the causal effect of work on health. Second, the project uses administrative data containing precise and reliable information on individuals’ labor market position and timing of transitions over time, as well as objective information about health (e.g. doctor and hospital visits, type of illness, treatment). Third, the data contain a universe of workers. This avoids problems with selective sampling and non-random nonresponse. Also, there are sufficiently large numbers of observations to get precise estimates.The results from this project will be important inputs in the current debate on whether or not working is harmful for health. Having credible estimates of the effect of work on health is essential not only because working is an important activity for many people, but also in the light of the political debate on changing the retirement age."