We use a randomized field experiment to measure the costs to smokers of decision errors identified in the behavioral economics literature, and incorporate the behavioral costs into improved estimates of the economic impact of tobacco control policies.