PURPOSE: Peer-mediated interventions are effective for improving social outcomes for children with ASD (Chang & Locke, 2016). However, parents often report difficulty identifying a typically developing (TD) peer who is willing to participate in an intervention for the child with ASD. For many families a TD sibling is a feasible option as the peer mediator (Ferraioli, Hansford, & Harris, 2012). However, little is known about the differences in play performance of the child with ASD when the play partner is a sibling vs a non-sibling. The Ultimate Guide to Play, Language and Friendship (PLF) is a peer-mediated intervention that has demonstrated effectiveness in improving play performance for 6- to 11-year old children with ASD using both TD sibling and non-sibling peers (Kent, Cordier, Joosten, Wilkes-Gillan, & Bundy, under review). The aim of this investigation was to compare the play of sibling and non-sibling peer play partners and examine the influence of the two types of play partners on the play performance of the child with ASD. DESIGN: Participants were 136 children aged between 6 and 11 years, (68 were diagnosed with ASD; 68 were TD playmates), and participated in a two-group parallel randomised controlled trial to test the effectiveness of the PLF. We engaged in a retrospective cohort study to investigate the difference in play performance of sibling (n=37) and non-sibling peer dyads (n=27). METHOD: An independent assessor scored using the Test of Playfulness (ToP; Skard & Bundy, 2008) from video recorded play sessions; we compared the play scores using an independent t-test. To identify mediating factors on play performance, we performed hierarchical regression analysis. RESULTS: Results revealed a significant difference in ToP scores for children with ASD when they played with a sibling vs a non-sibling play partner (ToP score of children with ASD in sibling dyads M=53.40, SD=12.14; in non-sibling peer dyads M=60.98, SD=12.23, t = -2.43, p=0.018). Individual item differences revealed that the social items 'support of playmate' and 'response to play cues' had the greatest difference, indicating that these need to be a focus of intervention for sibling dyads. CONCLUSION AND IMPACT STATEMENT: The findings revealed that the play performance of children with ASD is influenced significantly by their play partners. Practitioners need to consider these differences explicitly in intervention, particularly when, as is often the case, siblings are the play partners. Improved understanding of the difference in play performance will support practitioners to adapt interventions for children with ASD.