The longer-term problems of stroke are well documented but are often poorly addressed by community services. The Longer-Term Stroke (LoTS) care system of care aimed to address this gap through enhancing the practice of health care professionals termed ‘Stroke Care Coordinators’. The system was evaluated in a Randomised Controlled Trial (RCT) that measured patient and carer outcomes in comparison to usual care. These outcomes do not reveal how, why or to what extent the system enhanced the practice of the Stroke Care Coordinators. This study was designed to complement the RCT using a theory-driven approach to explore the implementation and the impact of the system at two community stroke services (both multidisciplinary teams). The theory of change (how and why the system was expected to work) was elicited for comparison against service practice. The study drew on the principles of realist evaluation, which hypothesise that successful outcomes (O) will be realised if appropriate ideas and opportunities (mechanisms (M)) are introduced into appropriate contexts (C). CMO propositions were drawn out from the wider theory for testing at the two research sites. Qualitative methods were employed for data collection including observations of service practice, interviews with stakeholder groups and a review of service documentation. The findings revealed that local facilitators and barriers shaped how the system was implemented, resulting in two distinct applications of the intervention that deviated from the theory of change. The extent to which the system enhanced the Stroke Care Coordinators’ practice was dependent on context e.g. their background and experience. Further to this, the system’s impact was mediated by the information and support available from within the multidisciplinary teams. For this reason, the extent to which the service enhancements realised in practice impacted on the patient and carer outcomes measured was hard to discern.