When is it permissible to rely on a proposition in practical reasoning? Standard answers to this question face serious challenges. This paper uses these challenges to motivate a certainty norm of practical reasoning. This norm holds that one is permitted to rely on p in practical reasoning if and only if p is epistemically certain. After developing and defending this norm, I consider its broader implications. Taking a certainty norm seriously calls into question traditional assumptions about the importance of belief and knowledge. In particular, it raises the possibility that many epistemological jobs that are usually assigned to belief and knowledge should be reallocated to two related but importantly different states: psychological and epistemic certainty.