Summary With increasing work-loads in anticoagulant clinics different methods of service delivery need evaluation. The quality of anticoagulant control achieved by a nurse-practitioner using a computer decision-support system (CDSS) was compared with that achieved by trainee doctors without CDSS. Eighty-one out-patients (group A, therapeutic range 2–3) and 96 out-patients (group B, therapeutic range 3–4.5) were randomized to management by a nurse-practitioner or by trainee doctors (clinicians). Thirty-seven patients in group A and 50 patients in group B were randomized to be managed by the nurse-practitioner. In group A, patients in the nurse-practitioner group spent a longer time in the therapeutic range than those in the clinician group (60.7% compared with 51.6%). Dose suggestion acceptance in the nurse-practitioner group (88%) was higher compared with agreement between the CDSS and the clinicians (60%). In group B, patients in the clinician group spent a slightly longer time in the therapeutic range (70% compared with 67.6%). Acceptance of dose suggestion was lower in the nurse-practitioner group (67%) compared with agreement between the CDSS and the clinicians (73%). In conclusion, the CDSS can improve the quality of control of warfarin therapy by a nurse-practitioner over that by trainee doctors for the therapeutic range 2–3. Similar quality of control is achieved for the therapeutic range 3–4.5. The CDSS may be used by nurse-practitioners to achieve safe and effective anticoagulation in hospital-based or out-reach anticoagulant clinics.