The use of cardiopulmonary bypass (CPB) during cardiac surgery is associated with a hemostatic defect, the hallmark of which is a markedly prolonged bleeding time. However, the nature of the putative platelet function defect is controversial. In this study, blood was analyzed at 10 time points before, during, and after CPB. We used a whole-blood flow cytometric assay to study platelet surface glycoproteins in (1) peripheral blood, (2) peripheral blood activated in vitro by either phorbol myristate acetate, the thromboxane (TX)A2 analog U46619, or a combination of adenosine diphosphate and epinephrine, and (3) the blood emerging from a bleeding-time wound (shed blood). Activation-dependent changes were detected by monoclonal antibodies directed against the glycoprotein (GP)Ib-IX and GPIIb-IIIa complexes and P-selectin. In addition, we measured plasma glycocalicin (a proteolytic fragment of GPIb) and shed-blood TXB2 (a stable breakdown product of TXA2). In shed blood emerging from a bleeding-time wound, the usual time-dependent increase in platelet surface P-selectin was absent during CPB, but returned to normal within 2 hours. This abnormality paralleled both the CPB-induced prolongation of the bleeding time and a CPB-induced marked reduction in shed-blood TXB2 generation. In contrast, there was no loss of platelet reactivity to in vitro agonists during or after CPB. In peripheral blood, platelet surface P-selectin was negligible at every time point, demonstrating that CPB resulted in a minimal number of circulating degranulated platelets. CPB did not change the platelet surface expression of GPIb in peripheral blood, as determined by the platelet binding of a panel of monoclonal antibodies, ristocetin-induced binding of von Willebrand factor, and a lack of increase in plasma glycocalicin. CPB did not change the platelet surface expression of the GPIIb-IIIa complex in peripheral blood, as determined by the platelet binding of fibrinogen and a panel of monoclonal antibodies. In summary, CPB resulted in (1) markedly deficient platelet reactivity in response to an in vivo wound, (2) normal platelet reactivity in vitro, (3) no loss of the platelet surface GPIb-IX and GPIIb-IIIa complexes, and (4) a minimal number of circulating degranulated platelets. These data suggest that the "platelet function defect" of CPB is not a defect intrinsic to the platelet, but is an extrinsic defect such as an in vivo lack of availability of platelet agonists. The near universal use of heparin during CPB is likely to contribute substantially to this defect via its inhibition of thrombin, the preeminent platelet activator.