This study utilises a novel approach to investigate the effectiveness of different learning modalities by combining video-based learning with eye-tracking. An excerpt taken from a vocational education instruction for car mechanics was videotaped using two different cameras: a standard 2D video camera and a professional 360° camera. The video recorded with the 2D camera was presented on a tablet, with a fixed angle, whereas the video recorded with the 360° camera was presented as non-interactive 3DoF virtual reality (nVR) environment using a head-mounted display. In both conditions, participants' fixation patterns were recorded and analysed in conjunction with a set of standardised questionnaires. Participants (N = 48) were randomly assigned to either the 2D-video group or the nVR group, with 23 participants in the 2D-video and 25 participants in the nVR group. The task of the participants in both groups was to watch the educational video while wearing an eye-tracker and then complete a standardised test on the presented content. The eye-tracking data indicated that participants in the nVR group showed longer total fixation durations on the instructor, but not other areas of interest, compared to the 2D video group. The standardised test indicated no differences in learning outcome between the groups. Implications from the current study as well as limitations and a outlook for further research will be discussed.