OBJECTIVE: Cost-effectiveness of safety measures in blood transfusion is not yet widely researched by pharmacoeconomic investigators. Only a few studies have been conducted, primarily in the US. Our objective is to review the literature on cost-effectiveness of interventions that enhance the safety of allogeneic (i.e. non-autologous) transfusion of whole blood and blood products. This review forms the basis for a European cost-effectiveness model for new technologies in transfusion safety. METHODS: We searched MEDLINE from 1990 onwards using the following search terms: "transfusion" in combination with "costs" or "cost-effectiveness". All cost-effectiveness analyses in the English language that use QALY's or life-years gained as primary health outcomes were included. RESULTS: Approximately 10 studies matched the selection criteria, all published in 1994 or later. The safety measures evaluated were donor screening, viral inactivation of frozen plasma, and single-donor versus random-donor platelet collection. Cost-effectiveness ranged from cost savings for screening donors for hepatitis C antibodies, to several million US$ per QALY gained as a result of screening donors for HIV using p-24 or nucleic-acid tests. Cost-effectiveness of viral inactivation of plasma and of shifting to single-donor platelet collection strongly depends on the patient population. For example, US$100,000-200,000 per QALY gained for coronary artery bypass graft and US$400,000-500,000 for cancer patients. CONCLUSIONS: In the USA and Canada thresholds for acceptable cost-effectiveness are US$50,000 and Can$20,000-100,000 (~US$13,000-67,000) per QALY gained. For the Netherlands, the threshold has recently been defined at Dfl40,000 per life-year gained (~EUR18,000, ~US$16,000). Preliminary results from our European model applied to the recently introduced nucleic-acid testing of Dutch donors for HIV indicate cost-effectiveness ratios of US$100,000 per life-year gained and above. Most cost-effectiveness ratios for interventions in transfusion safety are beyond currently accepted thresholds for cost-effectiveness. This may imply that different thresholds for cost-effectiveness apply when making decisions about the introduction of new technologies that enhance transfusion safety.