Blood Services, which, in the UK, spend over 0.5% of the NHS budget, are generally subject to quality, regulatory, economic and political authority. As only persons in good health should give blood, Services have refined donor selection criteria and aim to base them on evidence; but they also have to balance the number of donations collected with product demand. Applying selection criteria inevitably leads to deferrals, which donors experience very negatively. Compared with successful donors, even temporary deferrals reduce return rates significantly, especially of first attenders. In order to encourage donor return and sustain supplies, selection criteria should be optimal. However, a major tool for managing patients--evidence from randomized controlled trials (RCTs)--cannot apply to donor selection, so criteria have to be defined by alternatives, such as clinical studies, epidemiology and even what experts deem to be pathophysiologically feasible. The recommended volume of blood taken from each donor at each attendance (450 mL, which was based on old studies) was increased because of greater processing losses (buffy-coat derived platelets, leucofiltration etc.). Although faint rates and donation-induced iron depletion are reduced by lowering bleeding volume and bleeding less frequently, other optimizing strategies including iron supplementation have been trialled and could be enhanced by more RCTs. Better but more complex indicators of donor iron status than one-off Hb thresholds are possible. Regulators and decision-makers must encourage more studies. This review does not consider aphaeresis donors of blood components other than red cells in detail, or the prevention of transfusion-transmitted infections.