Hepatitis B virus (HBV) expresses co-terminal large (L), middle (M), and small (S) envelope proteins containing preS1/preS2/S, preS2/S, and S domain alone, respectively. S and preS1 domains mediate sequential virion attachment to heparan sulfate proteoglycans and sodium taurocholate cotransporting polypeptide (NTCP), respectively, which can be blocked by anti-S and anti-preS1 antibodies. How anti-preS2 antibodies neutralize HBV infectivity remains enigmatic. The late stage of chronic HBV infection often selects for mutated preS2 translation initiation codon to prevent M protein expression, or in-frame preS2 deletions to shorten both L and M proteins. When introduced to infectious clone of genotype C or D, both M-minus mutations and most 5' preS2 deletions sustained virion production. Such mutant progeny viral particles were infectious in NTCP-reconstituted HepG2 cells. Neutralization experiments were performed on the genotype D clone. Although remaining susceptible to anti-preS1 and anti-S neutralizing antibodies, M-minus mutants were only partially neutralized by two anti-preS2 antibodies tested while preS2 deletion mutants were resistant. By infection experiments using viral particles with lost versus increased M protein expression, or a neutralization escaping preS2 deletion only present on L or M protein, we found that both full-length L and M proteins contributed to virus neutralization by the two anti-preS2 antibodies. Thus, immune escape could be a driving force for the selection of M-minus mutations, and especially preS2 deletions. The fact that both L and M proteins could mediate neutralization by anti-preS2 antibodies may shed light on the underlying molecular mechanism.IMPORTANCEThe large (L), middle (M), and small (S) envelope proteins of hepatitis B virus (HBV) contain preS1/preS2/S, preS2/S, and S domain alone, respectively. The discovery of heparan sulfate proteoglycans and sodium taurocholate cotransporting polypeptide (NTCP) as the low- and high-affinity HBV receptors could explain neutralizing potential of anti-S and anti-preS1 antibodies, respectively, but how anti-preS2 neutralizing antibodies work remains enigmatic. In this study, we found two M-minus mutants in the context of genotype D partially escaped two anti-preS2 neutralizing antibodies in NTCP-reconstituted HepG2 cells, while several naturally occurring preS2 deletion mutants escaped both antibodies. By point mutations to eliminate or enhance M protein expression, and by introducing preS2 deletion selectively to L or M protein, we found binding of anti-preS2 antibodies to both L and M proteins contributed to neutralization of wild-type HBV infectivity. Our finding may shed light on the possible mechanism(s) whereby anti-preS2 antibodies neutralize HBV infectivity.