So far, most neutralizing antibodies recognize the RBD in the S protein S2 of CoVs. Compared with the high mutation rate in the S1 protein, S2 is much more conservative, thereby decreasing the off-target risk caused by amino acid replacement [62], and also bypassing the special epitopes that may cause ADE [63]. This means that the cocktail of monoclonal antibodies binding to different epitopes of RBD would be more desirable for therapeutic purposes [64]. For treatment, the monoclonal antibodies are from a human source or are humanized antibodies, isolated or generated with various approaches. For example, wild-type mice were immunized with soluble recombinant RBD containing the S protein. Then mouse antibodies were humanized and isolated, or transgenic mice were directly immunized, to express human versions of the antibodies [50,65,66]. However, direct cloning of single B cells from human survivors, used in combination with the phage-display antibody library, could provide authentic human antibodies. Until now, it should be noted that many neutralizing antibodies have been successfully discovered for treatment of SARS-CoV [67] and MERS-CoV infection [45,68,69]. These antibodies have all been described favorably in the literature [29,70,71]. A similar approach is known as single chain fragment variable (scFv) library screening, whereby the use of RBD as a bait protein allows some neutralizing antibodies to be screened out from non-immune humans [72,73].