PLATELET aggregation is mediated by at least three distinct mechanisms1,2. The first involves the release of ADP and is inhibited by its conversion to ATP by the combination of creatine phosphate and creatine phosphokinase (CP/CPK). The second is mediated by metabolites of arachidonic acid, particularly thromboxane A2 (TXA2), and is blocked by aspirin or indomethacin, inhibitors of the arachidonate cyclo-oxygenase pathway3. It has been postulated that a third mechanism must exist, as neither CP/CPK nor aspirin, alone or combined, can inhibit aggregation induced by high concentrations of thrombin or the calcium ionophore A23187 (refs 1, 2, 4). Antigenic challenge of IgE-sensitised basophils releases a platelet-activating, factor (PAF), probably a 1-lysophosphatidylcholine5. PAF has a potent action on rabbit6,7 and human8,9 platelet aggregation and release which is independent of the cyclo-oxygenase arachidonate pathways6,10,11. We have also obtained PAF from A23187-stimulated rat peritoneal12 and alveolar (J.B., B. Arnoux and D. Duval, in preparation) macrophages. Moreover, thrombin and ionophore-induced platelet aggregation and the accompanying stimulation of phospholipase A2 (refs 13, 14) are inhibited by the phospholipase inhibitor bromophenacyl bromide (ref. 15 and B.B.V., F. Fouque and M.C., in preparation), suggesting that the third mechanism of platelet aggregation might involve a lipid mediator. These findings prompted us to investigate whether platelets can form and release PAF in experimental conditions in which the ADP and TXA2 pathways are fully blocked. We report here that this is indeed the case, and suggest PAF as a likely candidate for mediating the ‘third pathway’ of platelet aggregation.