For more than 50 years now, Donnellan's distinction between referential and attributive descriptions has been discussed, criticized, supported and considered from many points of view. The facts have largely been agreed upon but the debate still revolves around how to account for them, especially for referring by a misdescription, i.e., a description that does not fit what it is being used to refer to, for example when I refer to the man near the window as the man drinking a martini, whereas he has a tonic water. Here, I offer a reconstruction of the issue and sketch a meaning as use framework (in fact, Donnellan distinguishes two uses of descriptions), within which I discuss an account for referential descriptions, including apparent ones. Before closing, I put forward a unitary account of the two uses of a description, whereby the descriptive condition is always relevant, while denying that a description in referential use operates in the same way as a proper name.