Introduction The recruitment of community health volunteers to support the delivery of health programmes is a well-established approach in many countries, particularly where health services are not readily available. However, studies on management of volunteers are scarce and current research on human resource management of volunteers faces methodological challenges. This paper presents the protocol of a realist evaluation that aims at identifying the factors influencing the performance of community health volunteers involved in the delivery of a Red Cross immunisation programme in Kampala (Uganda) with a specific focus on motivation. Methods and analysis The realist evaluation cycle structures the protocol. To develop the theoretical basis for the evaluation, the authors conducted interviews and reviewed the literature on community health volunteers’ performance, management and organisational behaviour. This led to the formulation of the initial programme theory, which links the intervention inputs (capacity-building strategies) to the expected outcomes (positive work behaviour) with mechanisms that point in the direction of drivers of motivation. The contextual elements include components such as organisational culture, resource availability, etc. A case study design will be adopted. We define a case as a Red Cross branch, run by a programme manager, and will select two cases at the district level in Kampala. Mixed methods will be used in data collection, including individual interviews of volunteers, participant observation and document review. The thematic analysis will be based on the initial programme theory and will seek for context-mechanism-outcome configurations. Findings from the two cases will be compared. Discussion We discuss the scope for applying realist evaluation and the methodological challenges we encountered in developing this protocol. Ethics and dissemination The study was approved by the Ethical Committee at Rennes University Hospital, France. Results will be published in scientific journals, and communicated to respondents and relevant institutions.