BackgroundCommunity-based peer support (CBPS) has been proposed as a potentially promising approach to improve health literacy (HL) and reduce health inequalities. Peer support, however, is described as a public health intervention in search of a theory, and as yet there are no systematic reviews exploring why or how peer support works to improve HL.ObjectiveTo undertake a participatory realist synthesis to develop a better understanding of the potential for CBPS to promote better HL and reduce health inequalities.Data sourcesQualitative evidence syntheses, conceptual reviews and primary studies evaluating peer-support programmes; related studies that informed theoretical or contextual elements of the studies of interest were included. We conducted searches covering 1975 to October 2011 across Scopus, Global Health (including MEDLINE), ProQuest Dissertations & Theses database (PQDT) [including the Education Resources Information Center (ERIC) and Social Work Abstracts], The King’s Fund Database and Web of Knowledge, and the Institute of Development Studies supplementary strategies were used for the identification of grey literature. We developed a new approach to searching called ‘cluster searching’, which uses a variety of search techniques to identify papers or other research outputs that relate to a single study.Study eligibility criteriaStudies written in English describing CBPS research/evaluation, and related papers describing theory, were included.Study appraisal and synthesis methodsStudies were selected on the basis of relevance in the first instance. We first analysed within-programme articulation of theory and appraised for coherence. Cross-programme analysis was used to configure relationships among context, mechanisms and outcomes. Patterns were then identified and compared with theories relevant to HL and health inequalities to produce a middle-range theory.ResultsThe synthesis indicated that organisations, researchers and health professionals that adopt an authoritarian design for peer-support programmes risk limiting the ability of peer supporters (PSs) to exercise autonomy and use their experiential knowledge to deliver culturally tailored support. Conversely, when organisations take a negotiated approach to codesigning programmes, PSs are enabled to establish meaningful relationships with people in socially vulnerable groups. CBPS is facilitated when organisations prioritise the importance of assessing community needs; investigate root causes of poor health and well-being; allow adequate time for development of relationships and connections; value experiential cultural knowledge; and share power and control during all stages of design and implementation. The theory now needs to be empirically tested via further primary research.LimitationsAnalysis and synthesis were challenged by a lack of explicit links between peer support for marginalised groups and health inequalities; explicitly stated programme theory; inconsistent reporting of context and mechanism; poor reporting of intermediate process outcomes; and the use of theories aimed at individual-level behaviour change for community-based interventions.ConclusionsPeer-support programmes have the potential to improve HL and reduce health inequalities but potential is dependent upon the surrounding equity context. More explicit empirical research is needed, which establishes clearer links between peer-supported HL and health inequalities.Study registrationThis study is registered as PROSPERO CRD42012002297.FundingThe National Institute for Health Research Public Health Research programme.