Research evidence does not necessarily translate into changed management for individual patients, but that may not mean that the evidence has been ignored. Drawing on accounts from general practitioners, we use a study of non-rheumatic atrial fibrillation (NRAF) to illuminate the processes by which practitioners became aware of and assimilated research evidence. We follow that with an account of how the evidence was incorporated into practice protocols for anticoagulation and then applied to a review of individual patients' records. Practitioners used a range of sources of evidence. They reported difficulties arising from their own skills and circumstances and from the perceived quality of the evidence. Creating a protocol involved overcoming problems of scheduling, resources and managing judgements about the value of the review process. In applying the protocol practitioners drew on their knowledge of the patients' preferences, circumstances and previous specialist consultations. As a result, practitioners made judgements that evidence, combined with prior experience, did not support the initiation of anticoagulation in 52% of an unselected primary care population with NRAF. Our findings have implications for evidence-based practice and for practitioner education.