The present paper assesses the causal effects of education on smoking and self-rated health in Canada. Education is instrumented using the changes in compulsory schooling laws. The sample is restricted to cohorts born between 1946 and 1964. The data are from the Canadian General Social Surveys of 1995 and 2016, allowing to observe the evolution of the gradient over the life course. The gradient is confirmed for smoking and self-rated health in correlational estimations, with stronger associations among males in later life. However, the IV regressions produce no evidence for a causal relationship.