Risk management concepts from the engineering literature suggest that a background risk of one in a million is the intuitive level of acceptable risk. That risk level is being approached by the risk of transfusion-associated malaria and HTV in Australia. However, the risk of transfusion error and the risk of transfusion-associated HBV is still unacceptable by this standard. The risk of transfusion-associated HIV is intermediate. It is likely that improvement in the risk of transfusion-associated disease will come primarily from the extended use of autologous transfusions and by technological solutions. By its very nature, donor assessment is an inefficient way of selecting safe donors, gives progressively diminishing returns and is prone to failure.