Background: Primary health care is changing as it responds to demographic shifts, technological changes and fiscal constraints. This, and predicted pressures on medical and nursing workforces, raises questions about staffing configurations. Physician assistants (PAs) are mid-level practitioners, trained in a medical model over 2 years at postgraduate level to work under a supervising doctor. A small number of general practices in England have employed PAs. Objective: To investigate the contribution of PAs to the delivery of patient care in primary care services in England. Design: A mixed-methods study conducted at macro, meso and micro organisational levels in two phases: (1) a rapid review, a scoping survey of key national and regional informants, a policy review, and a survey of PAs and (2) comparative case studies in 12 general practices (six employing PAs). The latter incorporated clinical record reviews, a patient satisfaction survey, video observations of consultations and interviews with patients and professionals. Results: The rapid review found 49 published studies, mainly from the USA, which showed increased numbers of PAs in general practice settings but weak evidence for impact on processes and patient outcomes. The scoping survey found mainly positive or neutral views about PAs, but there was no mention of their role in workforce policy and planning documents. The survey of PAs in primary care (n = 16) found that they were mainly deployed to provide same-day appointments. The comparative case studies found that physician assistants were consulted by a wide range of patients, but these patients tended to be younger, with less medically acute or complex problems than those consulting general practitioners (GPs). Patients reported high levels of satisfaction with both PAs and GPs. The majority were willing or very willing to consult a PA again but wanted choice in which type of professional they consulted. There was no significant difference between PAs and GPs in the primary outcome of patient reconsultation for the same problem within 2 weeks, investigations/tests ordered, referrals to secondary care or prescriptions issued. GPs, blinded to the type of clinician, judged the documented activities in the initial consultation of patients who reconsulted for the same problem to be appropriate in 80% (n = 223) PA and 50% (n = 252) GP records. PAs were judged to be competent and safe from observed consultations. The average consultation with a physician assistant is significantly longer than that with a GP: 5.8 minutes for patients of average age for this sample (38 years). Costs per consultation were £34.36 for GPs and £28.14 for PAs. Costs could not be apportioned to GPs for interruptions, supervision or training of PAs.Conclusions: PAs were found to be acceptable, effective and efficient in complementing the work of GPs. PAs can provide a flexible addition to the primary care workforce. They offer another labour pool to consider in health professional workforce and education planning at local, regional ...