Being able to follow the direction of another person’s line-of-sight facilitates social communication. To date, much research on the processes involved in social communication has been conducted using computer-based tasks that lack ecological validity. The current paradigm assesses how accurately participants can follow a social partner’s line-of-sight in a face-to-face scenario. In Study 1, autistic and neurotypical adults were asked to identify which location, on a grid of 36 potential locations, the experimenter was looking at on a series of discrete trials. All participants (both autistic and neurotypical) were able to effectively make line-of-sight judgements, scoring significantly above chance. Participants were also just as effective at making these judgements from either a brief, 1s, glance or from a prolonged, 5s, stare. However, at the group level, autistic participants were significantly less accurate than neurotypical participants overall. In Study 2, potential variation in performance along the broad autism phenotype was considered using the same paradigm. Bayesian analyses demonstrated that line-of-sight judgement accuracy was not related to the amount of autistic traits. Overall, these findings advance the understanding of the mechanistic processes of social communication in relation to autism and autistic traits in a face-to-face setting.Lay abstractIn order to effectively understand and consider what others are talking about, we sometimes need to follow their line-of-sight to the location at which they are looking, as this can provide important contextual information regarding what they are saying. If we are not able to follow other people’s line-of-sight, this could result in social communication difficulties. Here we tested how effectively autistic and neurotypical adults are at following a social partner’s line-of-sight during a face-to-face task. In a first study, completed by 14 autistic adult participants of average to above-average verbal ability and 14 neurotypical adult participants, we found that all participants were able to effectively follow the social partner’s line-of-sight. We also found that participants tended to be as effective at making these judgements from both a brief, 1s, glance or a long, 5s, stare. However, autistic adults were less accurate, on average, than neurotypical adults overall. In a second study, a separate group of 65 neurotypical adults completed the same line-of-sight judgement task to investigate whether task performance was related to individual variation in self-reported autistic traits. This found that the amount of self-reported autistic traits was not at all related to people’s ability to accurately make line-of-sight judgements. This research isolates and furthers our understanding of an important component part of the social communication process and assesses it in a real-world context.