We study intergenerational spillovers of environmental education using a randomized field experiment with 1,446 child–parent pairs in Patna, India, assigned to child-only, parent-only, joint, or control arms. Treating either children or parents raises the likelihood that the untreated household member chooses a delayed recycled certificate over an immediate standard one by 25 percentage points—spillovers on this incentivized behavior run symmetrically in both directions. Spillovers on beliefs and attitudes are asymmetric, however: children shift parents' views on climate change, but little spillover runs from parents to children on other measures. Joint participation does not outperform targeting children alone once child-to-parent spillovers are accounted for, suggesting that targeting children is a more scalable, cost-effective way to promote sustainable household behavior.