Recent findings from psychological studies have shown that emotional arousal improves human memory. However, more evidence is necessary if these results are generalisable to multimedia learning environments. Considering these needs, the study has the goal to examine the effect of emotional arousal on multimedia learning. Fifty-seven participants were presented with instructional materials with either high- or low-arousing words and pictures in an experimental study. The eye movements of participants were recorded while they studied the instructional materials to examine the online processes during learning. The results suggest that emotional arousal enhanced recall and transfer scores. The eye-tracking results demonstrate that emotional arousal attracted attention. The results of the mediation analysis suggest that fixation time on emotional pictures as an indicator of attention mediated the relationship between emotional arousal and learning outcomes. The findings show the importance of the guidance of attention by emotional multimedia elements for learning.