We develop and estimate a life-cycle model in a rational addiction framework where youth choose to smoke, attend school, work part-time, and consume while facing borrowing constraints. The model features multiple channels for studying the reciprocal causal effects of addiction and education. Variations in endowments and cigarette prices are sources of identification. We show that education causally reduces smoking. A counterfactual experiment finds that in absence of cigarettes, college attendance rises by three percentage points in the population. A practical alternative of 40% additional excise tax achieves similar results. Impacts vary substantially across persons of different cognitive and non-cognitive abilities.