Background: The recruitment of community health volunteers (CHVs) to support the delivery of health programmes is an established approach in underserved areas and in particular where there are health inequalities due to the scarcity of trained human resources. However, there is a dearth of evidence about what works to improve CHVs’ performance. This review aimed to synthesise existing literature to explain why, how and under which circumstances intervention approaches to improve the performance of CHVs are more likely to be successful.Methods: We performed a realist synthesis. We identified candidate theories related to our review questions, which then guided the selection, appraisal and analysis of primary studies. Publications of interest dating from 2008 to 2012 were identified by a systematic search in PubMed and IDEAS databases. We considered all study designs that examined the various aspects of CHV performance in the context of formal organisational settings to be eligible and excluded the studies that did not provide explanation about the performance of CHVs neither in the findings nor in the discussion part. Data were arranged according to their reference to context, interventions, outcomes and mechanisms in order to identify the interaction between them. The synthesis of data allowed us to determine explanatory patterns within or across the studies.Results: We identified broad intervention approaches within the 23 papers included in the review: positioning of the CHV within the community, establishment of clear roles, provision of skill-based and ongoing training, incentives, supervision and logistical support for task distribution and implementation. The findings provided information regarding which mechanisms (self-esteem, sense of duty, self-efficacy, sense of being fairly treated) to target when implementing such approaches, and which contextual factors (stable and supportive cultural, political and social context and intervention closely linked to local health services) create the most favourable conditions for these mechanisms to occur, ultimately contributing to CHVs’ better performance. Four main explanatory patterns around these mechanisms emerged as being fundamental to better performance.Conclusions: The patterns identified, combined with the designers’ and other stakeholders’ assumptions on how such interventions are expected to work, can be tested by empirical studies in order to provide useful information to be used by programme implementers, policymakers, donors and the community.