Long term institutional care in the UK is provided by care homes. Residents have prevalent cognitive impairment and disability, have multiple diagnoses and are subject to polypharmacy. Prevailing models of healthcare provision -ad hoc, reactive and co-ordinated by general practitioners -result in unacceptable variability of care. A number of innovative responses to improve health care for care homes have been commissioned. The organisation of health and social care in the UK is such that it is unlikely that a single solution to the problem of providing quality healthcare for care homes will be identified that can be used nationwide.Realist evaluation is a methodology which uses both qualitative and quantitative data to establish an in depth understanding of what works, for whom, in what settings. In this paper we describe a protocol for using realist evaluation to understand the context, mechanisms and outcomes which shape effective health care delivery to care home residents in the UK.By describing this novel approach, we hope to inform international discourse about research methodologies in long-term care settings internationally.