Background The uptake and use of clinical guidelines is often insufficient to change clinical behaviour and reduce variation in practice. As a consequence of diverse organisational contexts, the simple provision of guidelines cannot ensure fidelity or guarantee their use when making decisions. Implementation research in surgery has focused on understanding what evidence exists for clinical practice decisions but limits understanding to the technical, educational and accessibility issues. This research aims to identify where, when and how evidence and knowledge are used in orthopaedic decision-making and how variation in these factors contributes to different approaches to implementation of clinical guidance in practice. Methods We used in-depth case studies to examine guideline implementation in real-life surgical practice. We conducted comparative case studies in three English National Health Service hospitals over a 12-month period. Each in-depth case study consisted of a mix of qualitative methods including interviews, observations and document analysis. Data included field notes from observations of day-to-day practice, 64 interviews with NHS surgeons and staff and the collection of 121 supplementary documents. Results Case studies identified 17 sources of knowledge and evidence which influenced clinical decisions in elective orthopaedic surgery. A comparative analysis across cases revealed that each hospital had distinct approaches to decision-making. Decision-making is described as occurring as a result of how 17 types of knowledge and evidence were privileged and of how they interacted and changed in context. Guideline implementation was contingent and mediated through four distinct contextual levels. Implementation could be assessed for individual surgeons, groups of surgeons or the organisation as a whole, but it could also differ between these levels. Differences in how evidence and knowledge were used contributed to variations in practice from guidelines. Conclusion A range of complex and competing sources of evidence and knowledge exists which influence the working practices of healthcare professionals. The dynamic selection, combination and use of each type of knowledge and evidence influence the implementation and use of clinical guidance in practice. Clinical guidelines are a fundamental part of practice, but represent only one type of evidence influencing clinical decisions. In the orthopaedic speciality, other distinct sources of evidence and knowledge are selected and used which impact on how guidelines are implemented. New approaches to guideline implementation need to appreciate and incorporate this diverse range of knowledge and evidence which influences clinical decisions and to take account of the changing contexts in which decisions are made. Electronic supplementary material The online version of this article (10.1186/s13012-018-0771-4) contains supplementary material, which is available to authorized users.