Recent studies have claimed to show a significant causal impact of education on health status. Their empirical strategy usually relied on changes in compulsory schooling laws. Using a French longitudinal dataset, we focus on the effect of school leaving age on mortality at later ages. The two identifying shocks are the Zay and Berthoin reforms, which respectively raised the minimum school leaving age to 14 and 16 years. We implement a non-parametric regression discontinuity design, comparing cohorts born immediately before or after the reforms, and a parametric two-stage approach using information from a larger part of our sample. None of these approaches reveals a significant causality of education on health. Despite the fact that these reforms increased education levels, and that subsequent declines in mortality are observed, none of these declines appears to be significant. We conclude with a discussion on possible limitations of these two reforms as identifying devices.