Whether visual information belongs in second language (L2) listening tests has long been a subject for scholarly debate, with L2 learners' performance on and perceptions of video-based tests being the primary sources of evidence. The research into L2 teachers' perceptions, however, is scarce, as is the research into stakeholders' views of content visuals, such as a graph or diagram, in a listening assessment construct. This study sought to bridge these gaps by exploring stakeholders' perceptions of audio-only vs. video-based academic listening tests, the latter featuring a combination of context (e.g. the speaker's posture) and content (e.g. a graph) visuals. The questionnaire data from 143 English-as-a-second-or-foreign language (ESL/EFL) learners and 310 ESL/EFL teachers showed that both the learners and the teachers generally found video lectures to be less difficult, more motivating, more authentic, and suitable for high-stakes L2 academic listening tests. Although most stakeholders favored video-based lectures, wide variation among the learners' and teachers' perceptions suggests that some stakeholders had reservations about the role of videos in listening tests. These findings are discussed in the light of their implications for the assessment construct of L2 academic listening.