We re-examine the canonical question of Myerson and Satterthwaite (1983) whether two privately-informed parties, a buyer and a seller, can trade an indivisible good efficiently. We relax their assumption that utilities are quasi-linear; our main assumption instead is that the traded good is...