The present chapter attempts to resolve a puzzle about normative testimony. On the one hand, agents act on the advice of others, advice which purports to tell them what they have reason to do. When they do so, they can act for good reason. This thought, though, sits uneasily with another: that the mere fact that someone has advised a course of action is not itself a reason. An interesting view of reasons recently defended by Stephen Kearns and Daniel Star offers a resolution to the puzzle. On this view, reasons are evidence, specifically evidence concerning what one ought to do. If they are correct, then sufficiently good advice is, it turns out, itself a reason, and it is then no puzzle that an agent who acts on such advice is acting for a good reason. The chapter argues that this account of reasons is subject to counterexamples. There are reasons which are not evidence and evidence which is not a reason. This view of reasons cannot therefore hold the key to solving the puzzle of normative testimony. The solution to the puzzle of normative testimony lies instead in a more careful account of what it is to act for a reason. Such an account can preserve both the intuition that advice is not an independent reason and the thought that agents who act on sound advice act for good reason. The account also explains how agents can act for “elusive reasons.” These are facts which are reasons but which would not be if the agent became sufficiently aware of them.