The aim was to evaluate the efficacy and safety of two recombinant factor VIIa (rFVIIa) dose regimens for treating haemarthroses in patients with congenital haemophilia A or B and inhibitors. This was a multicentre, randomised, cross-over, double-blind trial. Patients were randomly allocated to treat a first joint bleeding episode with one 270 mu g/kg rFVIIa dose followed by two doses of placebo at 3-hour intervals and a second joint bleed with three single doses of 90 mu g/kg rFVIIa at 3-hour intervals, or vice versa. Efficacy was evaluated using a novel and robust treatment response-rating scale based on patient-assessment of pain and joint mobility. Outcome was rated at different timepoints, and an effective or ineffective treatment response was determined. Treatment '' preference '' was defined as effective treatment with one regimen and ineffective with the other. Patients with equally effective or ineffective treatments had no '' preference ''. Treatment was rated as effective for 65% of patients using the 270 mu g/kg dose versus 70% for the 90 mu g/kg x 3 regimen. An equal '' preference '' was noted for the two regimens (21% for each; p=0.637); most patients (58%) had no '' preference ''. 37/42 bleeding episodes (88%) were successfully treated with rFVIIa; additional haemostatic medications were administered for five episodes. No safety issues were identified. Administration of rFVIIa as a single 270 mu g/kg dose to treat haemarthroses in patients with haemophilia and inhibitors was at least as efficacious and safe as the 90 mu g/kg x 3 regimen.