The residual risk of transmitting viral infections by transfusion of screened blood is mainly linked to donations occurring in the window period. Using a mathematical model, a French study, including a 3 year period of blood donations (1994-1996), was performed by the Hepatitis Virus and Retrovirus working groups of the French Society of Blood Transfusion. The residual risk in France was estimated as follows: for HIV, 1 in 1 million (IC 95% = 1/10 million to 1/295,000), for HTLV 1 in 5 million (IC 95% = 0 to 1/625,000), for VHC 1 in 200,000 (IC 95% = 1/530,000 to 1/97,000), for VHB 1 in 180,000 (IC 95% = 1/560,000 to 1/66,000). Based on these estimations, the number of recipients possibly contaminated should have been seven for HIV, one or two for HTLV, 35 for VHC, 40 for VHB for a 3 year period. These 83 to 84 theoretically contaminated recipients are not in accordance with the six recipients contaminated by a blood product (two with HIV, one with VHC, three with VHB) that were reported to the National Haemovigilance Unit of the Agence Française du Sang. This observed difference may be explained by an overestimated calculated risk (the whole window period is not infectious), or by an underestimation of real post-transfusion contamination due to a lack of exhaustive notifications. However, a national database of post-tranfusion contamination is useful to approach the real residual risk. Therefore, efforts to continue the recipient follow-up have to be encouraged.