Understanding the underlying visual scanning patterns of individuals with autism spectrum disorder (ASD) during the processing of complex emotional scenes remains limited. This study compared the complex emotion recognition performance of adults with ASD (n = 23) and matched neurotypical participants (n = 25) using the Reading the Mind in Films Task. Behaviourally, both groups exhibited similar emotion recognition accuracy. Visual fixation time towards key social regions of each stimuli was examined via eye tracking. Individuals with ASD demonstrated significantly longer fixation time towards the non-social areas. No group differences were evident for the facial and body regions of all characters in the social scenes. The findings provide evidence of the heterogeneity associated with complex emotion processing in individuals with ASD.