OBJECTIVE--To elucidate the mechanism of the bleeding tendency observed in patients with aortic valve stenosis. DESIGN--A prospective study of high and low shear platelet function tests in vitro in normal controls compared with that in patients with severe aortic valve stenosis with a mean (SD) systolic gradient by Doppler of 75 (18) mm Hg before and at least 4 months after aortic valve replacement. SETTING--District general hospital. RESULTS--The patients showed reduced retention in the high shear platelet function tests. (a) Platelet retention in the filter test was 53.6 (12.6)% in patients with aortic valve stenosis and 84.8 (9.6)% in the controls (P < 0.001). (b) Retention in the glass bead column test was 49.8 (19.2) in the patients and 87.4 (8.7) in the controls (P < 0.001). (c) The standard bleeding time was longer in the patients (P < 0.06). Results of the high shear tests (a, b, and c) after aortic valve replacement were within the normal range. The platelet count was low but within the normal range before surgery and increased postoperatively (P < 0.01). There were no differences in the results of standard clotting tests, plasma and intraplatelet von Willebrand's factor, or in 15 platelet aggregation tests using five agonists between patients with aortic valve stenosis and controls. CONCLUSIONS--The high shear haemodynamics of aortic valve stenosis modify platelet function in vivo predisposing to a bleeding tendency. This abnormality of platelet function is detectable only in vitro using high shear tests. The abnormal function is reversed by aortic valve replacement. High shear forces in vitro activate and then inactivate platelets. By the same mechanisms aortic valve stenosis seems to lead to high shear damage in vivo, resulting in a clinically important bleeding tendency in some patients.