The antigenic site of dengue type 2 virus (DEN2)-neutralizing monoclonal antibody (mab) 3H5 was investigated by mutational analysis. Sequence comparisons indicated that much of the 12-amino-acid sequence extending from position 386 to 397 of the DEN2 envelope glycoprotein (E) previously thought to represent the DEN2-specific mab 3H5 binding site was also present in some dengue type 1, 3, or 4 virus strains. However, the region occupied by the Glu-Pro-Gly sequence at upstream positions 383 to 385 was completely conserved among DEN2 strains, but divergent in other serotype viruses, suggesting that this sequence might be part of the antigenic site of mab 3H5. We investigated this possibility by employing the previously constructed chimeric DEN2(PreM-E)/DEN4 cDNA clone to produce viable mutants bearing DEN2 PreM and E sequences that could be analyzed for binding to and neutralization by mab 3H5. We constructed 13 such DEN2 mutants that contained a single amino acid substitution in the region between positions 383 and 393 of DEN2 E. Each single substitution in the region spanning positions 386 through 393 of DEN2 yielded a virus that was as reactive with mab 3H5 as the parental chimeric virus. These results are consistent with the extent of sequence conservation in the region. In contrast, 5 of 6 mutants that sustained an amino acid substitution at position 383, 384, or 385 failed to react with mab 3H5 as detected by immunofluorescence assay and failed to be neutralized by the mab. Interestingly, each of the 5 mab-resistant DEN2 mutants also exhibited reduced mouse neurovirulence compared to parental chimeric DEN2 when inoculated intracerebrally. These observations suggest that the Glu-Pro-Gly sequence at positions 383-386 of the DEN2 E is a component of the site against which mab 3H5 is directed. In the recently determined three-dimensional structure of the related tick-borne encephalitis virus E, the Glu-Pro-Gly sequence would be located on the lateral surface of the immunoglobulin-like domain that is proposed to bind to the host cell receptor.