One key desideratum of a theory of blame is that it be able to explain why we typically have differing blaming responses in cases involving significant degrees of luck. T.M. Scanlon has proposed a relational account of blame, and he has argued that his account succeeds in this regard and that this success makes his view preferable to reactive attitude accounts of blame. In this paper, I aim to show that Scanlon's view is open to a different kind of luck-based objection. I then offer a way of understanding moral luck cases which allows for a plausible explanation of our differential blaming responses by appealing to the salience of certain relevant features of the action in question.