These conclusions refer to the four different types of platelet transfusion trial separately. Firstly, there is no evidence that a prophylactic platelet transfusion policy prevents bleeding. Two large trials comparing a therapeutic versus prophylactic platelet transfusion strategy, that have not yet been published, should provide important new data on this comparison. Secondly, there is no evidence, at the moment, to suggest a change from the current practice of using a platelet count of 10 x 10(9)/L. However, the evidence for a platelet count threshold of 10 x 10(9)/L being equivalent to 20 x 10(9)/L is not as definitive as it would first appear and further research is required. Thirdly, platelet dose does not affect the number of patients with significant bleeding, but whether it affects number of days each patient bleeds for is as yet undetermined. There is no evidence that platelet dose affects the incidence of WHO grade 4 bleeding.Prophylactic platelet transfusions were more effective than platelet-poor plasma at preventing bleeding.