The initiation and stabilization of oral anticoagulant therapy in hospitalized patients in a setting without specialized medical or pharmaceutical advice, was studied. In addition, potential risk factors for lack of stabilization were studied. All patients from three wards (orthopaedic surgery, general surgery and internal medicine) in two Dutch hospitals, who were started on oral anticoagulant therapy and who gave informed consent, were included in this three months prospective follow-up study. When a patient had two consecutive INR's within the range 2-3 during hospitalization (on day 6 or later), he was defined as stable. Stable and unstable patients were compared with respect to age, gender, quetelet index, length of hospital stay, indication for oral anticoagulant therapy, induction dosing schedule of oral anticoagulant therapy, prescribing physician, type of hospital (teaching or non-teaching), concurrently used drugs, concurrently used drugs known to potentially interact with oral anticoagulant therapy (drug-drug interactions that influence INR) and (co)morbidity. A total of 125 patients, who all used acenocoumarol as oral anticoagulant, were recruited in the study. The study population mainly comprised orthopaedic discharges on prophylactic oral anticoagulants. The mean length of hospital stay was 14.5 days (median 11.0, standard deviation (SD) 10.2) for the patients included in the study (patients with a short length of stay < 6 days were excluded from the study, because of the definition of stability). 43 patients (34%) became stable during hospitalization. The second INR within the range was reached after on average 11.1 days (median 10.0, SD 4.5). 18 different induction dosing schedules were used. Differences in apparent risk of INR instability were statistically associated with length of hospital stay (odds ratio (OR) 0.85, 95% confidence interval (CI) 0.78-0.92), concurrent use of musculoskeletal drugs, mainly NSAIDs, (OR 1.68, 95% CI 1.04-2.72) and two individual prescribing physicians (OR 6.61, 95% CI 1.47-29.82 for one physician and OR 0.23, 95% CI 0.06-0.99 for the other physician). This population has a high percentage of instability and reaches stability relatively late. The instability was associated with length of hospital stay, the concurrent use of musculoskeletal drugs (mostly NSAID's) and physician. Most of the unstable patients had INR's below therapeutic range, suggesting a conservative dosing habit. Part of the instability may also be due to the many different physicians who dose their own patients. Interventions to improve dosing may aid in better stabilization in hospitalized patients and thus in reduced length of hospital stay.