Objective:To compare the performance of V3-loop peptide enzyme immunoassay (PEIA) methodologies from four different laboratories for subtyping HIV-1, and to determine the causes for the lack of correlation between V3-loop PEIA serotyping and subtyping by sequencing.Materials and methods:Synthetic peptides derived from the amino-acid consensus sequences of the V3-loop of group M strains representing genetic subtypes A–F as well as reference strains were evaluated in PEIA by four different laboratories for their ability to accurately determine the subtype in a panel of 85 sera obtained from persons infected with known HIV-1 subtypes (28 subtype A, 34 subtype B, four subtype C, 10 subtype D, seven subtype F, one each of subtype H and G). Furthermore, the V3 loop of the corresponding virus was compared with the V3 loop of the peptides used in PEIA.Results:The correlation between HIV-1 subtyping by sequencing and V3-loop PEIA from the different laboratories varied considerably for the different HIV-1 subtypes: subtype A (46–68%), B (38–85%), C (75–100%), D (29–50%), and F (17–57%). A 70% agreement between PEIA and sequencing subtypes was observed for samples with the concordant presence of the same octameric sequences in the V3 loop of the virus and the V3 loop of the peptide used in PEIA; however, only 42% of specimens with different V3-loop octameric viral and peptide sequences yielded concordant results in V3-loop serotyping and genetic subtyping.Conclusion:Our results indicate that V3-loop PEIA methodologies used in different laboratories correlate poorly with genetic subtyping, and that their accuracy to predict HIV-1 subtypes in sera of Belgian individuals infected with different HIV-1 subtypes (A, B, C, D, F, G and H) vary considerably. The poor correlation between serotyping and genetic subtyping was partly due to the simultaneous occurrence of subtype-specific octameric sequences at the tip of the V3 loop of viruses belonging to different genetic subtypes.