We have previously shown that the platelet-aggregating activity of human MG-63 and HOS osteosarcoma cells depends at least in part upon tumor cell surface-associated thrombospondin, and suggested that platelet-osteosarcoma cell interactions could occur through interactions with specific platelet membrane receptors. In this study, the platelet-aggregating activity of MG-63 and HOS cells was studied by using a variety of platelet disorders. Both osteosarcoma cell lines induced a biphasic platelet aggregation response when added to normal platelet-rich plasma, while the second phase of aggregation was absent when added to gray platelets (deficiency in alpha-granule proteins) and to aspirin-treated platelets. Platelets from two unrelated patients with type I Glanzmann's thrombasthenia (deficiency in glycoprotein (GP) GPIIb/IIIa) did not aggregate at all with osteosarcoma cells. Using giant platelets from three patients with Bernard-Soulier syndrome (deficiency in GPIb/IX), the aggregation response induced by MG-63 and HOS cells was monophasic and reversible when compared to normal-sized platelets and to giant platelets from a patient with May-Hegglin anomaly (no membrane GP defect). Because GPIb serves as a receptor for von Willebrand factor during hemostasis, aggregation experiments were also conducted with the platelet-rich plasma of two patients with a low plasma von Willebrand factor concentration (type I von Willebrand's disease) before and after the infusion of deamino-D-arginine vasopressin. MG-63 and HOS cells induced biphasic platelet aggregation both before and after deamino-D-arginine vasopressin treatment, while the ristocetin-dependent binding of von Willebrand factor to platelets only occurred after deamino-D-arginine vasopressin treatment. Preincubation of normal platelet-rich plasma with monoclonal antibody SZ-2 directed against the von Willebrand binding domain of GPIb did not inhibit the platelet-aggregation activity of osteosarcoma cells, whereas anti-GPIb antibody SZ-2 did inhibit ristocetin-induced platelet agglutination. In addition, anti-GPIX antibodies did not affect platelet-osteosarcoma cell interactions. In conclusion, our data demonstrate that the first phase of the platelet-aggregating activity of human osteosarcoma cells is initiated by the interaction of these tumor cells with platelet membrane GPIIb/IIIa, whereas the second phase, even if plasma von Willebrand factor is deficient, involves platelet membrane GPIb and the participation of platelet alpha-granule proteins in membrane-mediated events, making aggregation irreversible.