The guided participation relationship between parents and children, whereby parents sensitively support the child’s emerging abilities, is central to child development. However, difficulties with intersubjective engagement characteristic of autism can make supportive interactions between parents and children more challenging. In this study, we compared dyads of parents and children with autism (n = 16) and dyads of parents and children with other developmental disabilities (n = 14) on two coding schemes, the Dyadic Coding Scales (DCS) and a Parent Scaffolding coding scheme. Consistent with expectations, the ASD group showed more difficulties in parent–child engagement on the DCS. Although autistic children were more likely to resist parental input on the Parent Scaffolding task, their parents were equally as sensitive as those in the developmental disabilities group. Child factors on the Parent Scaffolding coding scheme were associated with DCS scores. Half (n = 8) of the children with autism received an academic school year-long modified version of Relationship Development Intervention delivered in a preschool setting, and all children with autism (n = 16) were re-assessed at the end of the preschool year. The dyads who had received RDI showed improvements in parent–child engagement on the DCS, both compared to baseline and compared to the group who did not receive the intervention. No changes in Parent Scaffolding were found. The results are discussed in relation to intervention priorities for children with autism.