Five anti-subtype O specimens were tested by anti-HIV-1/2 screening and confirmatory assays. They can be divided into three specimens, reactive with all ELISAs, independent of the nature of the antigen (recombinant proteins or peptides) and test configuration (indirect ELISA or double antigen/sandwich ELISA). One specimen was not detected by one peptide based ELISA. One specimen was only recognized by two ELISAs and should be considered as a marker sample for the weakness of currently used ELISAs with anti-subtype O. Three different immunoblot assays available commercially detected two of the specimens with a major binding of gp160 and other viral bands, especially the integrase and reverse transcriptase. Another two specimens lacked reactivity with glycoproteins almost completely, but showed some staining with the enzymes of HIV, and would most probably be interpreted as indeterminate. The fifth specimen, which was also missed by most of the ELISAs, had very faint staining of the gp160 and a very weak staining of p24, and would most probably be interpreted as negative. Adaption of currently available tests to anti-subtype O is needed for the future reliability of anti-HIV diagnostic reagents.