As multimodal texts become ubiquitous in the digital age, analyzing how writers interact with readers via visuals is becoming increasingly important. Enlightened by Kress and van Leeuwen's (1996, 2001) semiotic work, D'Angelo (2010) extended Hyland's (2000) text-based metadiscourse model and proposed a new framework for analyzing the visual metadiscourse of multimodal texts. We used D'Angelo's (2016) visual metadiscourse framework to evaluate the comprehensibility and organization of the EFL students' infographics in this study. Two research questions guided our inquiry: (1) How do EFL students use diverse visual metadiscourse resources in their infographic posters? (2) In what ways, if any, does the use of visual interactive resources relate to the assessed quality of the infographics? We collected a total of 122 infographic posters and coded 5 categories of interactive resources for each poster (i.e., information value, framing, connective elements, graphic elements, and fonts), and then carried out frequency counts for each category and subcategories, taking all the posters into account. Moreover, we graded the EFL students' infographics according to a multi-traits grading rubric, and then conducted multiple regression analyses of their use of visual interactive resources and the scores for their infographics in the aspect of visual effects. The results show that the students used diverse visual interactive resources to guide their audience through the text. The assessed quality of the infographics was found to be significantly correlated with the use of graphic elements and fonts. This study sheds new light on digital multimodal pedagogy, particularly on the role of visual metadiscourse in teaching and assessing multimodal texts in language/writing classrooms.