ObjectivesWorldwide, emergency healthcare systems are under intense pressure from ever-increasing demand and evidence is urgently needed to understand how this can be safely managed. An estimated 10%–43% of emergency department patients could be treated by primary care services. In England, this has led to a policy proposal and £100 million of funding (US$130 million), for emergency departments to stream appropriate patients to a co-located primary care facility so they are ‘free to care for the sickest patients’. However, the research evidence to support this initiative is weak.DesignRapid realist literature review.SettingEmergency departments.Inclusion criteriaArticles describing general practitioners working in or alongside emergency departments.AimTo develop context-specific theories that explain how and why general practitioners working in or alongside emergency departments affect: patient flow; patient experience; patient safety and the wider healthcare system.ResultsNinety-six articles contributed data to theory development sourced from earlier systematic reviews, updated database searches (Medline, Embase, CINAHL, Cochrane DSR & CRCT, DARE, HTA Database, BSC, PsycINFO and SCOPUS) and citation tracking. We developed theories to explain: how staff interpret the streaming system; different roles general practitioners adopt in the emergency department setting (traditional, extended, gatekeeper or emergency clinician) and how these factors influence patient (experience and safety) and organisational (demand and cost-effectiveness) outcomes.ConclusionsMultiple factors influence the effectiveness of emergency department streaming to general practitioners; caution is needed in embedding the policy until further research and evaluation are available. Service models that encourage the traditional general practitioner approach may have shorter process times for non-urgent patients; however, there is little evidence that this frees up emergency department staff to care for the sickest patients. Distinct primary care services offering increased patient choice may result in provider-induced demand. Economic evaluation and safety requires further research.PROSPERO registration numberCRD42017069741.