There is increasing evidence that the sudden, unpredicatable event initiating myocardial infarction is fissuring of an atherosclerotic plaque. The resulting haemorrhage into the arterial wall produces obstructive platelet thrombi, just as arterial haemorrhages elsewhere produce haemostatic platelet plugs. It has been suggested that such platelet aggregation depends on ADP originating in red cells which are subjected to excessive haemodynamic stress at the site of haemorrhage. The release of ADP from red cells has been demonstrated in vitro in equivalent condtions of shear stress; and other mechansims, such as activation by collagen, cannot account for the rapidity with which the platelets react. One of us (G.V.R.B.) has suggested that drugs capable of counteracting haemolysis might diminish the activating effect of erythrocytes on platelets and so inhibit their aggregation as thrombi. Thus, chlorpromazine, added to human blood at concentrations which diminish haemoylsis but do not directly affect platelet aggregation, prolonged the 'bleeding time' from small holes in artificial vessels where extravasation is terminated, as in living arterioles, by aggregated platelets. The bleeding time was also prolonged by apyrase, consistent with the conclusion that the chlorpromazine acted through decreasing plasma ADP. We show here that this occurs through the anti-haemolytic action of chlorpromazine.