Background: Alcohol misuse is a wicked problem that may be best addressed by applying a complex systems perspective to the development and evaluation of alcohol interventions. Public health researchers have increasingly advocated this perspective, but the methods for complex systems process evaluations are under-developed. This thesis aims to develop and apply a framework for the application of a complex systems perspective to process evaluations of interventions to reduce alcohol-associated harms. Methods: The research involved 4 elements: i) a qualitative study involving interviews (n=30) and a focus group to evaluate the mechanisms by which the intervention ‘Reducing the Strength’ (RtS) may generate multi-level changes; ii) a scoping review of 87 primary studies and 3 systematic reviews to describe the scope of complex systems alcohol research; iii) a systematic review of 21 complex systems process evaluations and the development of a framework for qualitative process evaluation from a complex systems perspective; and iv) the application of this framework to evaluate the ‘Late Night Levy’ (LNL) using documentary analysis, interviews (n=21) and observations (35.5 hours). Findings: Alcohol interventions may generate multiple changes within and beyond the systems into which they are implemented. Alcohol research taking a complex systems perspective focuses on individual and local systems, with far less analysis of regional, national and international systems. Process evaluations from a complex systems perspective describe systems at a single timepoint, but utilise few complexity concepts to analyse system change. A two-phase process evaluation framework illustrates how to assess mechanisms of system change following intervention implementation. Applying the framework to evaluate the LNL demonstrated how the levy generated system changes which were both anticipated and unanticipated by system actors. Conclusion: The process evaluation framework can produce holistic appraisals of how interventions generate system changes across system levels; evaluators should further apply and refine the framework.