Constant infusion of factor VIII (FVIII) into patients with haemophilia A after major surgery has been recommended as optimal treatment to avoid peaks and valleys in the circulating levels of FVIII and to allow the use of much lower doses of FVIII than are historically required. One of our young patients with severe (< 0.01 U/ml FVIII) haemophilia suffered a subdural haematoma for which he received treatment with 815190 recombinant FVIII (rFVIII) units over a period of 52d. 2 weeks after admission, because of low FVIII levels and the presence of FVIII inhibitors, the infusion rate was increased to > 100 U/kg/h for 14d. During this time the FVIII level fluctuated between 0.6 and 4.2 U/ml. For some period it was not possible to detect ristocetin co-factor activity in this patient's plasma and the von Willebrand factor (VWF) level and VWF multimer pattern resembled those of a patient with von Willebrand's disease. Subsequently, when the rFVIII dose was increased 2-fold, this was not reflected by the plasma level of FVIII although antibodies were not detected. The data suggest that the prolonged infusion of very high levels of rFVIII which is deficient in von Willebrand factor can result in depletion of VWF from existing stores, producing a laboratory picture which is consistent with the diagnosis of von Willebrand's disease. Further, in the absence of complexing with VWF, FVIII appears to be cleared from the circulation at an increased rate. This is expensive and potentially compromising. Therefore, when administering very high doses of FVIII concentrates devoid of VWF for prolonged periods of time, ristocetin cofactor and VWF levels should be monitored.