The paper explores Wittgenstein’s notion of grammar in the sense of a discipline or an activity, as opposed to the object sense of the term (grammar as a body of rules for the use of a language). I argue that the Wittgensteinian activity of grammar consists in giving expression to rules of our language use. It differs from the traditional grammarian’s activity not only in focusing on a different type of rules, but also in that it does not aim at (...) an explicit and exhaustive treatment of a specific domain of language. Instead, Wittgenstein conceives its goal as therapeutic: the dissolution of particular philosophical problems. Further, I attempt to reconcile his seemingly contradictory remarks on the character of grammatical statements, defining the senses in which they respectively can, and cannot, be considered descriptive assertions. I confront G. P. Baker’s and P. M. S. Hacker’s conceptions of the Wittgensteinian grammatical activity and I argue in favour of the former. Finally, I critically examine N. Garver’s claim that Wittgenstein, in his later conception of philosophy as grammar, succeeded in formulating a successfully self-referential criterion of philosophical critique. I also argue that grammatical activity, despite Wittgenstein’s overt commitment, is in fact not the only method of his later philosophy. (shrink)