Background: Interest in how to implement evidence-based practices into routine health care has never been greater. Primary care faces challenges in managing the increasing burden of chronic disease in an ageing population. Reliable prescriptions for translating knowledge into practice, however, remain elusive, despite intense research and publication activity. This study seeks to explore this dilemma in general practice by challenging the current way of thinking about healthcare improvement and asking what can be learned by looking at change through a complexity lens. Methods. This paper reports the local level of an embedded case study of organisational change for better chronic illness care over more than a decade. We used interviews, document review and direct observation to explore how improved chronic illness care developed in one practice. This formed a critical case to compare, using pattern matching logic, to the common prescription for local implementation of best evidence and a rival explanation drawn from complexity sciences interpreted through modern sociology and psychology. Results: The practice changed continuously over more than a decade to deliver better chronic illness care in line with research findings and policy initiatives - re-designing care processes, developing community linkages, supporting patient self-management, using guidelines and clinical information systems, and integrating nurses into the practice team. None of these improvements was designed and implemented according to an explicit plan in response to a documented gap in chronic disease care. The process that led to high quality chronic illness care exhibited clear complexity elements of co-evolution, non-linearity, self-organisation, emergence and edge of chaos dynamics in a network of agents and relationships where a stable yet evolving way of organizing emerged from local level communicative interaction, power relating and values based choices. Conclusions: The current discourse of implementation science as planned system change did not match organisational reality in this critical case of improvement in general practice. Complexity concepts translated in human terms as complex responsive processes of relating fit the pattern of change more accurately. They do not provide just another fashionable blueprint for change but inform how researchers, policymakers and providers participate in improving healthcare.