In this contribution, I deal firstly with the problem of whether ‘religion’ can actually be defined. My answer is twofold. Firstly, that such a definition must indeed be deemed to be extremely un-like¬ly, if not downright impossible. Secondly, however, that definition also has more modest uses which may turn definitions of religion, that have shed this universalist ambition, into quite useful tools in the academic study of religions. In the second section, I shall address the question of why, if a (...) definition of religion turned out to be merely a useful research tool, one should bother to define ‘religion’ at all. My answer will be that one may indeed well dispense with it, but that, despite its very modest usefulness, it would still be unwise to do so. In my third section, I shall discuss these modest uses of definitions of religion, as well as their strategic implications. In the fourth and last section, I shall discuss the operational, or instrumental, definition of ‘religion’ which I have developed for my particular line of studies as an illustration of the pur¬poses which a definition of religion may serve in the academic study of religions. (shrink)