The last decade has seen substantial improvements in the provision of safe ‘infection-free’ blood for patients in many countries. This has resulted from the combined effects of better donor education, selection, testing and exclusion processes. The residual risk of infection with HBsAg attributable to laboratory error is less than 0.1/100 000 (1/750 000). The risk for HIV remains to be quantified but may approach this figure. With such low risks it will be difficult to provide statistical evidence that further improvements in the process of selection, testing or exclusion will have an impact on reducing risk from either agent. Over the same time, less progress has been made reducing risk to the recipient attributable to problems in the clinical supply process, i.e. getting the right blood, to the right person, at the right place, at the right time. Uniform definitions of terminology defining blood donation characteristics, together with the sharing of performance data are essential if we are to make national and international comparison of the risks that patients face when they receive a transfusion. Equally, the use of agreed definitions, and the sharing of data on the clinical outcomes of transfusion can provide the key to better prescribing based on evidence of actual risks and benefits.