AbstractThis paper defines the megaproject selection problem, discusses how it arises, and suggests some rules for improving the political decision processes inevitably involved in choosing between competing megaprojects. Any and all ex ante estimates about the cost-benefits of megaprojects depend on the judgments and forecasts of experts. However, to improve megaproject selection, the paper suggests forcing cost considerations early, widening the mix of experts used, taking note of predictive track record in selecting experts, convening experts more than once given flexible and modular designs, and reducing private knowledge and information races among US scientific cadres so as to increase international risk and cost sharing.