As soon as we consider promising from what seems to be the most natural point of view – namely a moral point of view – it is immediately seen as an act by which one undertakes a commitment: the obligation for the promiser to perform what she promised to do. Herbert Hart thus says: ‘Promises constitute the obvious case of moral obligation. […] When we promise we make use of specified procedures to change the moral situation; in lawyer's language we exercise a “power” conferred by rules to change moral relations.’1 Thus, when I promise to help someone, I create (for myself) the moral obligation to help this person. It is shown by the fact that one often tells children it is a bad thing not to keep their promises. Now the speech-act character of the promise is supposed to rely on this ‘creation’ of duty. That is to say that a promise consists in acting by committing oneself to perform what one said one would perform. The active aspect of a promise would thus lie in the taken commitment – a commitment which, at first glance, seems to be of a moral kind. Contrary to the one who merely expresses the intention to perform something, the one who promises does take the obligation to perform what she commits to; the one who promised cannot change her mind. Somehow she is bound to perform what she promised to do, as if the promise itself had a constraining force (or efficiency). But how is such a thing possible? To put the question the way Warnock once did: ‘[Why] (or how) does a promise induce, in the promise-giver, an “extra” sense that he ought to do the thing specified in the promise? How does that work? What is in promising, what could it be, that mysteriously generates a special sort of commitment to acting in a special way?’1 And, we may add, is this special sort of commitment a moral one?2 Is it because it is a(n) (intrinsically) moral (speech) act that a promise does generate a commitment? That is to say: is it because a promise belongs to the domain of morals that it is a speech act? In this paper, I shall address these questions by first examining the accepted conception of speech act theory (mainly John R. Searle's one) and, then, by criticizing its intentional account of the promise's efficiency (namely its ‘illocutionary force’). I will thus show (in a Humean and Austinian vein) that only a conventionalist conception can account for the special efficiency (a normative efficiency) of a promise (as a speech act). I will conclude that this efficiency is not intrinsically moral, even if the choice of using such a speech act pertains to a deliberate, and as such a moral, choice. Promising is not in itself moral but to make a promise – to play the game of promising – is, in our society, morally binding because it can be, and usually is, a moral practice.