Health systems in Australia and worldwide are increasingly expected to conduct research and quality improvement activities in addition to delivering clinical care and training health professionals. This study aims to inform a research impact evaluation at a regional Australian Hospital and Health Service by developing a programme theory showing how research investment is expected to have impact.This qualitative study, representing the first phase of a larger mixed methods research impact evaluation at the Townsville Hospital and Health Service (THHS), adopts a realist-informed design involving the development of a programme theory. Data were obtained between February and May 2019 from strategic documentation and interviews with six current and former health service executives and senior employees. Inductive themes were integrated into a conceptual framework to visually represent the programme theory.Research at THHS has developed organically as the service has matured into a regional tertiary referral service serving a diverse rural and remote population across northern Queensland. Throughout this journey, individual THHS leaders often adopted a research development mantle despite disincentives arising from a performance-driven reporting and activity-based funding service context. Impact expectations from research investment at THHS were identified in the categories of enhanced research activity and capacity among clinicians, and improved clinical practice, health workforce capability and stability, and patient and population health. Seven contextual factors were identified as potential enablers or obstacles to these impact expectations and ambitions.By identifying both relevant impact types and key contextual factors, this study offers programme theory to inform a planned research impact evaluation at THHS. The conceptual framework may be useful in other regionally based health service settings. More broadly, there are opportunities for future research to test and refine hybrid versions of linear and realist research impact evaluation models that combine resource-intensive, theory-driven approaches with policy practicality.