We know our own intentional actions in a distinctively first-personal way. Many accounts of knowledge of intentionally doing something, A, assume that grounds for the knowledge would have to establish or indicate that it is true that one is intentionally doing A. In this paper, I argue against this assumption, showing how it entails being in a Moore-paradoxical situation. I argue that if knowledge of intentionally doing A were such that grounds for it must be truth-indicating, then one could always (...) wonder, when doing A, whether A is for one a goal. However, just as wondering whether p is true is incompatible with thinking that one believes p, so wondering whether A is for one a goal is incompatible with thinking that one is intentionally doing A. We must allow, then, that one’s knowledge of intentionally doing A is itself a representation of A as a goal to be accomplished, apt to be grounded by reasons for doing A. I show that the first-personal character of knowledge of acting intentionally resides in its being practical rather than theoretical. (shrink)