Research abounds on deploying interventions to support peer interactions between children on the autism spectrum (AS) and neurotypical children [1‐2]. Technological advancements such as eye tracking devices have provided ample affordances to the field of autism research in terms of facilitating the collection of insightful data related to the gaze-based interaction behaviour of children on the AS [3]. Yet, further empirical understanding on how children interact within inclusive educational contexts is needed [4]. Addressing this gap, we present our research study set in a naturalistic inclusive school environment in which small groups of three children interact playing a boardgame, Alias, while wearing mobile eye tracking glasses. Alias is a word game during which a person guesses the word that another person is explaining by giving them hints and tips. In our study, we analysed the game play by focusing on 3 roles: an explainer, as the person giving clues for a guesser to guess what the word was, while an observer watched the interactions. 25 children (age 10‐12 years, 4th ‐ 5th grade, 15 males) from 3 different schools in Finland participated in the study. For data analysis, 2 categories were used: children with no diagnosis (n=19) and children on the autism spectrum (AS) (n=6). 43 data collection sessions of about 45 minutes each were carried out. Each session was set so that a small inclusive group of 3 children, each wearing mobile eye tracking glasses (Tobii Glasses 2), engaged for 10 minutes in a ludative task planned by the researchers and for 25 minutes in a curricular task planned by the teacher. Here we present the analysis of 9 sessions during which Alias was used. The mobile eye tracking glasses data collected was coded using Tobii Pro Lab. We analysed the children’s gaze behaviour from the perspective of the role taken, focusing on the proportion of time that the children spent looking at three areas of interest: other children’s faces, the task (i.e., the table where the game cards were placed) and the surrounding environment area. The statistical analysis of the coded data (i.e., areas of interests) was carried out using SPSS, and within-group comparisons were conducted with related samples Friedman’s two-way analysis of variance by ranks, with Dunn’s pairwise post hoc test with Bonferroni correction. Results show promising insights on the gaze-based interactions behaviour of children on the AS when compared with neurotypical peers, including a perceived sensitivity to the requirements of the game roles that they play. Our work contributes towards empirical research implementations of mobile eye tracking glasses to understand natural peer interactions in inclusive educational environments. The insights obtained could support teachers’ classroom practices to foster children’s collaborative work in inclusive educational contexts.