Résumé : Transferable utility (TU) is a widely used assumption in economics. In this paper, we weaken the TU property to the setting where distinct Pareto frontiers have empty intersections. We call this the no-intersection property (NIP). We show that the NIP is strictly weaker than TU, but still maintains several desirable properties. We discuss the NIP property in relation to several models where TU has turned out to be a key assumption: models of assortative matching, the Coase theorem and Becker's Rotten Kid theorem. We also investigate classes of utility functions for which theNIP holds uniformly.