Concerned with the question of how, as agents, we should take into account evidence when thinking about our future actions. Sometimes we promise and resolve to do things that we have evidence is difficult for us to do. Should we believe that we will follow through, or believe that there is a good chance that we won't? If we believe the former, we seem to be irrational since we believe against the evidence. yet if we believe the latter, we seem to be insincere since we can't sincerely say that we will follow through. Hence, it seems, our promise or resolution must be improper. To meet this challenge, the author considers and rejects a number of responses, before defending a solution inspired by the Kantian tradition and by Sartre in particular: as agents, we have a distinct view of what we will do. If something is up to us, we can decide what to do, rather than predict what we will do. But the reasons in light of which a decision is rational are not the same as the reasons in light of which a prediction is rational. That is why, provided it is important to us to do something, we can rationally believe that we will do it - even if our belief goes against the evidence.