Accounting for complexity is now a feature of health interventions research, but it is unclear how this might best be accomplished. As the number of methodologies to account for complexity expands, developing a coherent approach to intervention research has become more urgent and yet more difficult. This thesis aimed to address this challenge by examining methodologies used to design and evaluate complex interventions in global health. Four areas considered central to complex interventions research were explored – intervention design, evaluation of outcomes, assessment of causal mechanisms, and evaluation of context. In each of these areas, a different mixed method, statistical, or qualitative methodological approach was employed following available guidance. Data were drawn from the design and evaluation of the PRIME intervention, a complex health service intervention to improve care for malaria at health centres in rural Uganda. Conceptual and methodological challenges were encountered in each area of investigation. Opportunities for improving each methodological application are suggested alongside an overall recommendation for greater reflection on, and reporting of, the processes and investments necessary for conducting complex interventions research. Additionally, the evidence produced in each area of investigation revealed different, partial and incommensurable accounts of the intervention and its effects. This draws attention to the challenges that can arise when seeking to combine evidence of ‘what works’ with evidence from methodologies that employ different approaches to understanding how interventions are taken up and produce effects. Approaches to accounting for complexity in intervention research need to evolve from focusing on the narrow question of ‘what works’ towards emphasising a more dynamic and multi-perspective question of ‘what happens’. Such an approach may be particularly useful for understanding the multiple and varied effects of complex interventions and their role in improving health and wellbeing.