Background: Collaboration between parents and speech and language therapists (SLTs) is seen as a key element in family-centred models. Collaboration can have positive impacts on parental and children's outcomes. However, collaborative practice has not been well described and researched in speech and language therapy for children and may not be easy to achieve. It is important that we gain a deeper understanding of collaborative practice with parents, how it can be achieved and how it can impact on outcomes. This understanding could support practitioners in daily practice with regard to achieving collaborative practice with parents in different contexts. Aims: To set a research agenda on collaborative practice between parents and SLTs in order to generate evidence regarding what works, how, for whom, in what circumstances and to what extent. Methods & Procedures: A realist evaluation approach was used to make explicit what collaborative practice with parents entails. The steps suggested by the RAMESES II project were used to draft a preliminary programme theory about collaborative practice between parents and SLTs. This process generates explicit hypotheses which form a potential research agenda. Discussion & Conclusions: A preliminary programme theory of collaborative practice with parents was drafted using a realist approach. Potential contextual factors (C), mechanisms (M) and outcomes (O) were presented which could be configured into causal mechanisms to help explain what works for whom in what circumstances. CMO configurations were drafted, based on the relevant literature, which serve as exemplars to illustrate how this methodology could be used. In order to debate, test and expand our hypothesized programme theory for collaborative practice with parents, further testing against a broader literature is required alongside research to explore the functionality of the configurations across contexts. This paper highlights the importance of further research on collaborative practice with parents and the potential value of realist evaluation methodology.