People increasingly turn to online social networks forinformation and debate. This means that the structures andproperties of these networks, and the information theypropagate, play crucial roles in the development of socialbeliefs, attitudes, and morals. Recently, research has shownthat the presence of specific language drives the diffusion ofmoral messages, regardless of the informational quality, in aphenomenon dubbed moral contagion (Brady et al., 2017).Due to the widespread attention and implications of suchfindings for science and society, we investigate the presenceof moral contagion across six sets of data that capture thecommunications of naturally-occurring networks on Twitter.Across a large corpus of diverse tweets (n = 525,229), we findmoral contagion to be an inconsistent and often absentphenomenon that does not effectively predict messagediffusion. The implications and reasons for this finding arediscussed.