This paper concerns the semantic branch of meta-ethics, and examines a version of so called semantic pluralism advocated by Michael Gill. Briefly put, Gill suggests that ordinary people’s usage of moral terms is rather messy in the sense that the meaning of moral terms can vary not only between different people, but also for one and the same person in different contexts. Such variability in word-meaning is explained by his assumption that people’s meta-ethical commitments are part of their moral thought and language, which is to say that their meta-ethical commitments have implications for the meaning of moral terms. In this paper I pursue two objectives. The first is exegetical and aim to clarify how Gill’s semantic pluralism in general, and his Indeterminacy- and Variability theses in particular, are intended to be understood—specifically in relation to the cognitivist/non-cognitivist debate. The second objective is argumentative. I first present and evaluate an objection to Gill’s semantic pluralism from Walter Sinnott-Armstrong who argue that Gill’s Variability thesis implies that interlocutors with different meta-ethical commitments are talking past each other rather than having genuine first order moral agreements and disagreements. I then argue that a similar problem occurs also for certain second order moral disagreements, particularly those in which moral terms are used rather than mentioned. I then argue that this is problematic not only on independent grounds, but also because it is inconsistent with the very assumptions Gill makes to support his view. My argument therefore appears forceful by Gill’s own lights. Combined, Sinnott-Armstrong’s objection and my own leads me to conclude that Gill’s semantic pluralism does not look promising.