Moral error theorists hold that morality is deeply mistaken, thus raising the question of whether and how moral judgments and utterances should continue to be employed. Proposals include simply abolishing morality (Richard Garner), adopting some revisionary fictionalist stance toward morality (Richard Joyce), and conserving moral judgments and utterances unchanged (Jonas Olson). I defend a fourth proposal, namely revisionary moral expressivism, which recommends replacing cognitivist moral judgments and utterances with non-cognitivist ones. Given that non-cognitivist attitudes are not truth apt, revisionary expressivism does not involve moral error. Moreover, revisionary expressivism has the theoretical resources to retain many of the useful features of morality, such as moral motivation, moral disagreement, and moral reasoning. Revisionary expressivism fares better than the three major alternatives in both avoiding moral error and preserving these useful features of morality. I also show how this position differs from the “revolutionary expressivism” of Sebastian Köhler and Michael Ridge.