Aims-To determine the costs and effectiveness of an anticoagulant nurse specialist service compared with a conventional consultant service based on two hospital sites in northwest Hertfordshire. Methods-Sequential design comparing retrospectively the conduct and outcomes of a consultant service with a nurse specialist service over two six month periods. In each of the six month study periods, all new patients consecutively referred for anticoagulation over a three month period (group A) at the start of each study period and a random selection of patients who had already been attending the anticoagulant service for one year or more (group B) were included in the study. Group A patients were followed for up to three months and group B patients for six months. The main outcome measures were costs of service provision and effectiveness. Costs included those for the use of the anticoagulant service, those related to general practitioner (GP) visits and hospitalisations, and running costs (staff time, laboratory tests, patient transport). Measures of effectiveness were the mean proportion of time patients spend in the therapeutic range, the number of drugs being taken that could interact adversely and/or inhibit haemostatic function, and patient and GP satisfaction with service provision. Results-In the consultant service, for group A there were more patients aged 66-75 years (p = 0.004) and fewer patients aged more than 76 years (p = 0.001); and for group B, there were fewer patients on anticoagulation for cardiac conditions (p = 0.001), but more on anticoagulation for thromboembolic conditions (p = 0.02) than in the nurse specialist service. The clinic running costs of the nurse specialist service were (L)4.99 per attendance, compared with (L)4.75 in the consultant service. Including all other costs related to treatment, there was no statistically significant difference in cost per patient. There was no significant difference in the proportion of time patients spent in the therapeutic range between the consultant service and the nurse specialist service. In the nurse specialist service, fewer patients in group A were taking drugs that could interact adversely and/or inhibit haemostatic function (p = 0.01) and mole patients were satisfied with service provision (p = 0.04) compared with the consultant service. There was no significant variation in GP satisfaction between the two services. Conclusion-In the provision of outpatient anticoagulation, the nurse specialist service was no more expensive than the consultant service and, using our primary outcome, at least as effective. The nurse specialist service has some clear advantages compared to the consultant service: provision of domicilliary care for housebound patients, fewer new patients taking drugs that could interact adversely and/or inhibit haemostatic function patients, it is preferred by newly referred patients to the consultant service, and it is as acceptable to their GPs.