While speakers’ wording, prosody and gestures may affect perceivers’ liking of speakers, few studies investigate how teachers’ multimodal cues jointly impact students’ evaluation of second language (L2) teaching. We extracted 54 videos of vocabulary instruction delivered by four female L2 English teachers, varying features of prosody, linguistics, and gestures. 156 university students randomly watched 12 videos and rated their liking of each vocabulary teaching. Prosodic (speaking rate, mean pitch), linguistic (utterance length, question rate, total words) and gestural cues (iconic, beat) of videos were coded and analysed as predictors, while controlling for different teachers, teachers’ dressing formality, students’ working memory, English proficiency, and familiarity with the target vocabulary. Results showed that better working memory, higher English proficiency, and prior knowledge of the target word were positive predictors of students’ liking of teaching. Teachers using longer utterances, asking more questions tended to be less liked by students. Furthermore, male students significantly preferred teaching with a slower speaking rate, lower mean pitch, higher iconic gesture rate but lower beat rate, and more formal teacher attire. However, these effects were not significant for female students. In conclusion, teachers’ multimodal cues influence students’ liking of L2 teaching, with implications for education practice.