Realist evaluation, a methodology for exploring generative causation within complex health interventions to understand ‘how, why and for whom’ programmes work, is experiencing a surge of interest. Trends indicate that the proliferation in the use of this methodology also applies to research in low- and middle-income countries (LMICs). The value of using realist evaluation for project evaluation is also being noticed by non-governmental organisations (NGOs) and other programme implementers within such contexts. Yet, there is limited exploration of the use of realist evaluations in LMICs, especially their use by foreign researchers. This paper draws on the author’s experience of conducting two realist evaluations across three different sub-Saharan African settings: Mundemu, Tanzania; Kabale, Uganda and Marsabit, Kenya. The realist evaluations were used as an operations research methodology to study two NGO community health programmes. This paper highlights four main challenges experienced by the author throughout the methodological process: (1) power imbalances prevalent during realist interviews, (2) working through translation and what this means for identfying Context-Mechanism-Outcome Configurations, (3) limited contextual familiarity and being an ‘engaged researcher’ and (4) the use or dependence on ‘WEIRD’ theories (i.e. theories based on the study of Western, Educated, Industrialized, Rich, Democratic people) within testing and refinement. Realist evaluation’s enticing and straightforward slogan of finding ‘what works, for whom and why’ is in contrast to the complexity of the methodology used to generate these results (and often to the results themselves). Striking a balance between theory and pragmatism, while adhering to realist ontological underpinnings of generative causation and retroduction, is no easy task. This paper concludes by providing concrete recommendations for those who want to undertake a realist evaluation, with particular attention to cross-cultural settings, in light of the aforementioned challenges. In doing so, it aims to foster improved methodological rigour and help those engaging in this research methodology to work towards more appropriate and contextually relevant findings.