This paper interprets accidents occurring on the way to and from work as negative health shocks to identify the causal effect of health on labor market outcomes. We argue that in our sample of exactly matched treated and control workers, these health shocks are quasi-randomly assigned. A fixed-effects difference-in-differences approach estimates a negative and persistent effect on subsequent employment and income. After initial periods with a higher incidence of sick leave, treated workers are more likely unemployed, and a growing share of them leaves the labor market via disability retirement. Those treated workers, who manage to stay in employment, incur persistent income losses. The effects are stronger for sub-groups of workers who are typically less attached to the labor market.