This article aims to initiate exploration into the multimodal academic discourse of micro-lessons in the blended-learning context. It develops a multimodal framework of the image-text relations in micro-lessons from Visual Grammar (Kress& van Leeuwen, 1996) and Systemic-Functional image-text relations theory (Martinec et al., 2005). The model is then used to establish evaluation criteria for communication effectiveness of two award-winning micro-lessons. Two principles are derived from the framework: (a) an equal status of image and text; (b) varied and appropriate logic-semantic relations between visual symbols. Through a detailed multimodal intersemotic analysis of the distinct micro-lesson videos, the genres and emphasis of each micro-lesson appear more evident. It is also found that the styles of these micro-lesson videos vary according to disciplines. Therefore, it is proposed that this multimodal framework should be useful for categorizing different genres of micro-lesson videos and giving sound evaluations to the communication effectiveness of most micro-lesson videos. Further studies on the audience’s perception of these videos are suggested.