Experiments were undertaken to test a new hypothesis for the mechanism underlying the Révész effect. The hypothesis proposes that lethally irradiated (LI) tumour cells enhance the take probability of a small number of transplanted viable (V) tumour cells mixed with them by exerting a thromboplastic effect at the site of injection; local fibrin formation prevents emigration of V cells from the site or secures their survival there. The evidence presented to support this hypothesis is as follows: in the case of 3 isogeneically transplanted tumours, admixed particulate brain extract simulated the effect of LI cells in increasing the take probability of V cells; brain extract simulated the effect of LI cells in greatly delaying the disappearance of (125)IUdR-labelled viable carcinoma cells from the injection site; V cells acquired a raised take probability by their incorporation in fibrin clots; it was confirmed that admixed erythrocytes increased the take probability of V cells; using a newly devised microscopical test for detection of the thromboplastic activity of individual cells, it was found that cell death was almost always required for the display of such activity; lymphocytes and bone marrow cells, ineffective in enhancing the take of V cells, were almost totally devoid of thromboplastic activity. Possible explanations are given for failure of a fibrinogen depleting agent, ancrod (Arvin) to inhibit the Révész effect when administered to recipients. It is concluded that the evidence strongly supports the hypothesis presented whilst seriously weakening the long-standing theories that admixed LI cells act by provision of nutrients or by local quenching of postulated immune reactivity.