Social media enhances information exchange efficiency, yet it also facilitates the dissemination of rumors which can cause harmful consequences for companies and the society. Previous research mainly focuses on identifying online rumor messages and examining individual rumoring behavior; however, little is known about what makes some online rumor messages disseminate more than others. In this study, the effects of rumor topic-congruence and rumor message features, including cognitive appeals (i.e. information unambiguity) and affective appeals (i.e. emotional words and imagery words), on online rumor message dissemination are examined. To test the hypotheses, data were collected from Twitter and labelled via sentiment analysis and manual coding. This study contributes to previous literature by extending the conceptualization of information ambiguity and understanding how rumor topic-congruence moderates the effects of cognitive appeals and affective appeals on rumor message dissemination.