Background/aim: Children with ASD are known to have lower play skills than their typically developing peers. However, the play skills of children with ASD are rarely investigated using observational measures in the context of their everyday peer-to-peer play interactions. To explore the play skills of children with ASD and their aged matched classmates during a peer-to-peer play interaction. Methods: Using convenience sampling, four children with ASD (5–11 years) attending mainstream schools were recruited for this multiple case design study. Each child with ASD was paired with one of their aged matched typically developing classmates. Children's play skills were measured using the Test of Playfulness (ToP). Additional case data were collected through teacher-reported social skills and behaviours. Rasch analysis was utilised to convert raw ToP scores into an interval level overall score for each child. Children's individual ToP item scores, social skills and behaviours are presented by case. Results: The two children with ASD who had the highest ToP scores, also had the highest teacher-reported social skills. All children with ASD had greatest difficulty on ToP items reflecting suspension of reality and framing. Two children with ASD had higher ToP scores than their classmate. In these two cases, the classmates had similar play skills of children with ASD. Conclusion: The play skills of children with ASD varied by case. Across the cases, teacher-reported social skills, classmate age and existence of friendship between children were all factors observed to influence play. These findings require replication and investigation in larger scale studies.