The only certain physiological function of platelets is their aggregation in injured vessel walls as haemostatic plugs. The association of thrombocytopenia with petechial haemorrhages suggests that platelets are somehow required for the functional integrity of small vessels, but no mechanism has yet been established. The pathological aggregation of platelets as thrombi in atherosclerotic arteries is commonly, if not always, initiated by haemorrhage. In artificial vessels, platelets tend to aggregate on the walls wherever blood flow is non-laminar. The mural aggregation of platelets is not prevented by unphysiologically high wall-shear forces. The facts suggest, on the contrary, that the process depends in some way on abnormal haemodynamic conditions. This contribution is mainly concerned with questions about how haemodynamic conditions in and around vascular leaks affect arriving platelets that aggregate there, and about the chemical agents responsible for making the platelets reactive. The effects of these agents are known mainly from in vitro experiments in which aggregation can be quantitatively correlated with biochemical effects by simple and reproducible methods; the relevance to their reactions in haemostasis and thrombosis is uncertain. It is difficult to devise quantitative methods for analysing these processes in vivo because of the very low concentrations at which endogenous agents can activate platelets and haemostatic factors in the plasma; the rapidity with which platelets aggregate in a damaged blood vessel; and the complexity and inconstancy of the haemodynamic situation. All these facts must be accounted for in hypotheses of haemostasis. New experimental approaches towards analysing the haemostatic mechanism in vivo are described.