The well-known substantialist-'Frazerian'-definitions of magic as distinct from religion by its immediate and individual goals, the concomitant manipulative and coercive attitude, the instrumental and mechanical type of action etc., have been under attack for more than half a century. Anthropologists in particular have argued that no meaningful contrast between religion and magic can be gained from this approach and that our notion 'magic' is a modern-western biased construct which does not fit representations of other cultures. Consequently, in the view of some of them, the term 'magic' should be altogether avoided. Furthermore, with respect to the ancient and early modern world, in which the opposition religion-magic is supposed to have originated, it is argued that magic and religion function exclusively as value-judgments, terms indicating 'magic' being exploited to stigmatize illegitimate or undesired (religious) behaviour of socially or culturally deviant groups. In the present article it is argued that-although admittedly this functionalist approach has yielded remarkable and lasting results-rejection of the term 'magic' will soon turn out to be unworkable and, in fact, is putting the cart before the horse. From an etic point of view-which in the view of the author is the only possible way to conduct scholarly discourse-it will be impossible to do cultural research without the aid of heuristic instruments such as-at least broad, polythetic or prototypical-definitions. And, if possible at all, it would be utterly unpractical to completely eliminate religion as one of the obvious models of contrast. This position is substantiated with some practical instances from the Graeco-Roman world. It is shown that, at least in the context of (magical) curse-tablets and-related but clearly distinct-(religious) prayers for justice or vengeance, the ancient authors were clearly aware of the very same distinctions modern people normally associate with the notions of magic and religion.