Severe childhood malarial anemia is commonly treated using blood transfusion. Although transfusion may decrease short-term mortality, the risk of human immunodeficiency virus (HIV) transmission is considerable in Africa. We constructed a decision tree to weigh the short-term mortality benefit of transfusion against HIV infection risk. Probability estimates were derived from published studies. The base-case was a two-year-old child with a 13.5% mortality risk to be transfused with screened or unscreened blood (1% or 13% HIV contamination risk, respectively), with reduction of mortality to 5.5% by transfusion (odds ratio=2.7), and a 2.4% risk of fatal transfusion complications. A sensitivity analysis was performed to assess the influence of variation in these estimates. If a child developed acquired immunodeficiency syndrome, survival was weighed as one-tenth of normal survival. For the base-case, we found that transfusion with screened blood provided a survival benefit of 5%. In contrast, transfusion with unscreened blood decreased survival by 2%. Patients with a mortality risk < 5% derived no benefit from a transfusion with screened blood. Other important factors for the benefit of transfusion were the effectiveness of transfusion in reducing mortality and the risk of blood contamination. A blood transfusion was clearly beneficial if the mortality risk was high and the risk of contamination was low. Our findings can be used as a basis for a clinical transfusion policy that limits transfusions to situations in which they are likely to be beneficial. This will in turn optimize child survival and prevent unnecessary exposure of low risk children to the transfusion risks.