On average, people who retire earlier tend to die sooner. This is partly because poor health is an important reason for early retirement. But does retirement affect mortality? A number of studies have investigated the relationship between retirement and mortality. While the majority of studies find earlier retirement to be associated with higher mortality, there is no consensus about the causal relationship between retirement and mortality. In this thesis, I first investigate the association between retirement and mortality in two full Norwegian birth cohorts (1906 and 1907). Event history analyses show that those who have retired have a higher mortality risk than those who are the same age but have not retired. This difference increases with time since retirement, even when disability pensioners are excluded. I then turn to a pension reform that came into effect in 1973. This reform lowered the eligibility age for old age pensions in the national insurance scheme from 70 to 67 years. I treat this reform as a natural experiment whereby some birth cohorts were allowed to retire up to three years earlier than other cohorts. This allows me to estimate the effect of being eligible for retirement and the effect of actually retiring earlier. For these analyses, I sample five full Norwegian birth cohorts born between 1902 and 1906. I then exploit the exogenous variation in retirement eligibility age that stems from this reform by employing difference-in-differences, individual fixed effects and instrumental variables methods. The main advantage of these approaches is that the estimates can be given a causal interpretation. There are several other advantages to studying the 1973 reform. More or less the entire mortality history of the affected cohorts can be studied, and the reform affected most of the working population. Additionally, mortality was higher in the 1970s among people in their late sixties than among younger groups affected by later reforms. The results show that retirement increased mortality in 1970s Norway. Retiring one year earlier increased the probability of dying before age 80 by 1.5 percentage points among men and 0.5 percentage points among women. Among men, the effect was larger among those with low education, while the opposite was true for women. These effects appear to be driven by a relatively short-term increase in mortality following retirement. In sum, these results are in line with the view of retirement as a stressful life event and with theories predicting retirement to have short-term negative health consequences. The results can thus be taken to support the notion that postponing retirement may entail some health benefits.