Philosophers are by now familiar with “the” paradox of supererogation. This paradox arises out of the idea that it can never be permissible to do something morally inferior to another available option, yet acts of supererogation seem to presuppose this. This paradox is not our topic in this paper. We mention it only to set it to one side and explain our subtitle. In this paper we introduce and explore another paradox of supererogation, one which also deserves serious philosophical attention. (...) People who perform paradigmatic acts of supererogation very often claim and believe that their acts were obligatory. Plausibly, this is simply a mistake insofar as the actions really are “above and beyond the call of duty,” as common sense would have it. The fact that moral heroes tend to view their actions in this apparently mistaken way is puzzling in itself, and we might learn something interesting about the moral psychology of such individuals if we could explain this tendency. However, this puzzling aspect of the moral psychology of moral heroes is also the chief ingredient in a deeper puzzle, one perhaps more worthy of the title “paradox.” In this paper we present and try to resolve this paradox. The paradox arises when we combine our initial observation about the moral psychology of moral heroes with three plausible claims about how these cases compare with one in which the agent realizes her act is “above and beyond.” The first of these three additional claims is that the agent who mistakenly claims that the act is obligatory is no less virtuous than someone who performs such an act whilst correctly judging it to be obligatory. The second is that the agent who makes such a mistake would display more moral wisdom if she judged the act to be supererogatory. The third is that there is no other relevant difference between the two agents. These three claims, together with a plausible principle about the way in which the virtues work, give rise to a paradox. We consider several ways in which this paradox might be resolved. We argue that the most plausible resolution is to reject the claim that there is no other relevant difference between the two agents. More specifically, we argue that a relevant difference is that the agent who makes this mistake does so because of the depth of their commitment to certain moral values, and that this is itself an important moral virtue: moral depth. (shrink)