Using human monoclonal antibodies (HuMAbs) r(1)-447 (L-736,523) and 19b to the V3 region of HIV-1 gp120, we have explored epitope presentation on V3-peptides and on the corresponding gp120 proteins. HuMAb r(1)-447 binds strongly to the MN and SF-2 peptides and gp120 proteins. In contrast, while this HuMAb binds equally avidly to both the HxB2 and the BRU/BH10 peptides, it binds but weakly to the HxB2 V3 loop on gp120 and fails to bind at all to BH10 gp120. Thus, the solid-phase peptide binding assay can falsely predict reactivity of an MAb with a gp120 protein. Conversely, HuMAb 19b fails to bind to a peptide from the V3 loop of HIV-1 AD-6 in solid-phase assays, but binds to the same peptide in solution and also to AD-6 gp120. Thus, the solid-phase peptide binding assay can fail to predict reactivity of an MAb with a gp120 protein. Furthermore, serum antibodies from individual AD-6 do not react well with the AD-6 V3-peptide in a solid-phase assay, but react strongly with the corresponding MN V3-peptide. On the basis of peptide binding assays, we had assumed that the AD-6 virus was "MN-like" with a prototypic North American/European subtype B GPGR motif at the crown of the V3 loop. However, direct sequencing demonstrates that the AD-6 V3 loop contains a variant GPGK motif. This highlights a limitation of V3-peptide-based assays for serotyping viruses.