Griffithsin (GRFT) is a lectin that has been shown to inhibit HIV infection by binding to high mannose glycan structures on the surface of gp120, and it is among the most potent HIV entry inhibitors reported so far. However, important biochemical details on the antiviral mechanism of GRFT action remain unexplored. In order to understand the role of the three individual carbohydrate-binding sites (CBS) in GRFT, mutations were made at each site (D30A, D70A, and D112A), and the resulting mutants were investigated. NMR studies revealed that each GRFT variant was folded but showed significant peak movement on the carbohydrate-binding face of the protein. The wild-type and each point mutant protein appeared as tight dimers with a K(d) below 4.2 μM. Mutation of any individual CBS on GRFT reduced binding of the protein to mannose, and ELISA assays revealed a partial loss of ability of each GRFT point mutant to bind gp120, with a near-complete loss of binding by the triple mutant D30A/D70A/D112A GRFT. A more quantitative surface plasmon resonance (SPR) examination showed a rather small loss of binding to gp120 for the individual GRFT point mutants (K(D): 123 to 245 pM range versus 73 pM for wild-type GRFT), but dramatic loss of the triple mutant to bind gp120 derived from R5 and X4 strains (K(D) > 12 nM). In contrast to the 2- to 3-fold loss of binding to gp120, the single CBS point mutants of GRFT were significantly less able to inhibit viral infection, exhibiting a 26- to 1900-fold loss of potency, while the triple mutant was at least 875-fold less effective against HIV-1 infection. The disparity between HIV-1 gp120 binding ability and HIV inhibitory potency for these GRFT variants indicates that gp120 binding and virus neutralization do not necessarily correlate, and suggests a mechanism that is not based on simple gp120 binding.