This paper revisits one of the rare success stories in global environmental cooperation: the Montreal Protocol and the phase-out of ozone-depleting substances. I show that the protocol increased science and innovation on alternatives to ozone-depleting substances and argue that agreements can indeed be useful in solving global public goods problems. This contrasts with game-theoretical predictions that agreements occur only when costs to the players are low and with the often-heard narrative that substitutes were readily available. I reconcile theory and empirics by discussing the role of induced innovation in models of environmental agreements.