We examine the robustness of previous research finding increased diffusion of Twitter messages ("tweets") containing moral rhetoric. We use a distributed language model to examine the moral language used by U.S. political elites in two corpora of tweets: one from 2016 presidential candidates Hillary Clinton and Donald Trump, and one from U.S. Members of Congress. Consistent with previous research, we find greater diffusion for tweets containing moral rhetoric, but this is qualified by moral language valence and elite ideology. For both presidential candidates and Members of Congress, negative moral language is associated with increased message diffusion. Positive moral language is not associated with diffusion for presidential candidates and is negatively associated with diffusion for Members of Congress. In both data sets, the relationship between negative moral language and message diffusion is stronger for liberals than conservatives. (PsycInfo Database Record (c) 2022 APA, all rights reserved).