Background Emergency care is facing a steadily rising demand. In response, hospitals have implemented new models of care that locate general practitioners in or alongside the emergency department. Objectives We aimed to explore the effects of general practitioners working in or alongside the emergency department on patient care, the primary care and acute hospital team, and the wider system, as well as to determine the differential effects of different service models. Design This was a mixed-methods study in three work packages. Work package A classified current models of general practitioners working in or alongside the emergency department in England. We interviewed national and local leaders, staff and patients to identify the hypotheses underpinning these services. Work package B used a retrospective analysis of routinely available data. Outcome measures included waiting times, admission rates, reattendances, mortality and the number of patient attendances. We explored potential cost savings. Work package C was a detailed mixed-methods case study in 10 sites. We collected and synthesised qualitative and quantitative data from non-participant observations, interviews and a workforce survey. Patients and the public were involved throughout the development, delivery and dissemination of the study. Results High-level goals were shared between national policy-makers and local leads; however, there was disagreement about the anticipated effects. We identified eight domains of influence: performance against the 4-hour target, use of investigations, hospital admissions, patient outcome and experience, service access, workforce recruitment and retention, workforce behaviour and experience, and resource use. General practitioners working in or alongside the emergency department were associated with a very slight reduction in the rate of reattendance within 7 days; however, the clinical significance of this was judged to be negligible. For all other indicators, there was no effect on performance or outcomes. However, there was a substantial degree of heterogeneity in these findings. This is explained by the considerable variation observed in our case study sites, and the sensitivity of service implementation to local factors. The effects on the workforce were complex; they were often positive for emergency department doctors and general practitioners, but less so for nursing staff. The patient-streaming process generated stress and conflict for emergency department nurses and general practitioners. Patients and carers were understanding of general practitioners working in or alongside the emergency department. We found no evidence that staff concerns regarding the potential to create additional demand were justified. Any possible cost savings associated with reduced reattendances were heavily outweighed by the cost of the service. Limitations The reliability of our data sources varied and we were unable to complete our quantitative analysis entirely as planned. Participation in interviews and at case study sites was voluntary. Conclusions Service implementation was highly subject to local context and micro-level influences. Key success factors were interprofessional working, staffing and training, streaming, and infrastructure and support. Future work Further research should study the longer-term effects of these services, clinician attitudes to risk and the implementation of streaming. Additional work should also examine the system effects of national policy initiatives, develop methodologies to support rapid service evaluation and study the relationship between primary and secondary care. Trial registration This trial is registered as ISRCTN51780222. Funding This project was funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme and will be published in full in Health and Social Care Delivery Research; Vol. 10, No. 30. See the NIHR Journals Library website for further project information.