Objective To evaluate the implementation of a primary care, nurse-led, near patient anticoagulant monitoring service. Design Action research workshops, supported by questionnaires and clinical audit, to define the strengths and weaknesses of the service and the effectiveness of the computerised decision support system used to set the dosage of anticoagulant and time interval to the next appointment. Setting 13 general practices that implemented anticoagulant monitoring in a primary care organisation in south east England. Participants 18 practice nurses, 72% of whom had over 20 years' clinical experience; the universitybased investigators and managers from the primary care organisation. Main outcome measure The nurses felt that the patients preferred the practice-based service, finding it more personal and accessible. However, circumstances arose where the nurse's intuition had to override the software's advice. The nurses found it stressful when they were unclear whether their decision making represented acceptable variation or dangerous practice. An audit tool was developed to measure the extent to which there was variation from the software's recommendation, and patterns of variation emerged. Most evident was that nurses responded to uncertainty by practising cautiously, shortening the interval until the next visit and slightly reducing the recommended dose of warfarin. Conclusions The group, by sharing their experiences through a structured series of workshops, developed an understanding of when it might be appropriate to vary from the decision support software's recommendations and how this could be audited. The technological solution modelled on hospital practice proved hard to implement in primary care.Keywords (MeSH terms): anticoagulants, computerassisted, decision making, family practice, medical informatics, medical informatics applications, warfarin S de Lusignan, A Singleton and S Wells 28 Near patient testing allows monitoring to be provided outside the hospital setting. Traditionally patients had to travel to the local hospital for testing, where junior doctors adjusted their warfarin dose without the benefit of CDSS. 6,7 The combination of near patient testing and CDSS allows warfarin testing to take place in local clinics or in patients' homes, with the service provided by nurses or other non-medical staff. Where comparison has been made between nurses using computerised decision support and junior doctors using clinical judgement, the nurses achieved better control for patients with an INR target ratio of 2-3, and as good control as the doctors when target INR is 3-4.5. 8 A primary care organisation (PCO) recognised the potential benefits of a practice-based anticoagulation service and implemented it across 13 practices, deciding that practice nurses would be the appropriate professional group to deliver the service. They put on a training course comprising the relevant biomedicine: how anticoagulants work, the patients who need it and for how long, interactions and safety issues. The p...