Background: We have undertaken a systematically searched literature review using a realist logic of analysis to help synthesise the diverse range of literature available on hospice at home services. Aim: To find out in the existing literature what features of hospice at home models work best, for whom and under what circumstances. Design: A realist logic of analysis was applied to synthesise the evidence focusing on mechanisms by which an intervention worked (or did not work). An initial programme theory was developed using the National Association for Hospice at Home standards, Normalisation Process Theory and through refinement using stakeholder engagement. Data sources: PubMed, Science Direct, AMED, BNI, CINAHL, EMBASE, Health Business Elite, HMIC, Medline, PsychINFO, SCOPUS, Web of Science, DARE, Google Scholar, NHS Evidence, NIHR CRN portfolio database, NIHR journal library of funded studies, including searches on websites of relevant professional bodies (August 2014, June 2017, June 2019). Results: Forty-nine papers were reviewed, of which 34 contributed evidence to at least one of the eight theory areas: marketing and referral, sustainable funding model, service responsiveness and availability, criteria for service admission, knowledge and skills of care providers, integration and coordination, anticipatory care, support directed at carers. Conclusions: Our literature review showed how it was possible to develop a coherent framework and test it against 34 published papers and abstracts. Central to this review was theory building, and as further evidence emerges, our programme theories can be refined and tested against any new empirical evidence.