A comparative evaluation of the sensitivity of anti-HIV screening assays has been recently performed with a selected panel including all the actually known difficulties of HIV serology: Very recent seroconversions (per-seroconversions) and recent HIV1/M seroconversions, HIV1/M seropositive with a subtype different from B, HIV1 group 0 and HIV2. The criteria of sensitivity were to recognize as positive all the seropositive samples and at least 2 of the 8 per-seroconversions. The 18 combined ELISA fulfilled these criteria, except for one assay with a deficient HIV2 sensitivity. However, all these assays do not have the same level of sensitivity. The most important differences were observed with the pre-seroconversion samples, the number of positive results on such samples varying from 2 to 8. The capacity of a screening assay to early detect an HIV-infected subject is a supplementary criterium to retain, particularly in blood transfusion screening, in order to shorten the "window" period.