Membrane fusion is the main pathway for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) to invade host cells. Harringtonine (HT), derived from cephalotaxus fortunei Hook. f., has been recognized as an effective antagonist of SARS-CoV-2. It can directly block the active binding of spike (S) protein to host angiotensin converting enzyme 2 (ACE2), as well as hinder the enzymolysis of transmembrane serine proteases 2 (TMPRSS2). This study examined the potential of HT metabolites, 5'-de-O-methylharringtonine and cephalotaxine, as the membrane fusion inhibitors for SARS-CoV-2. 5'-De-O-methylharringtonine was synthesized and subsequently characterized by high resolution mass spectrometry and nuclear magnetic resonance to be structurally consistent, with a purity of 92.677% determined by reverse phase high performance liquid chromatography. Both 5'-de-O-methylharringtonine and cephalotaxine can specifically bind to SARS-CoV-2 S protein and TMPRSS2 using cell membrane chromatography. They can form hydrogen bonds with key sites that correlated highly with the enhanced binding affinity of SARS-CoV-2 and its variants to ACE2 or nafamostat to TMPRSS2. Moreover, 5'-de-O-methylharringtonine and cephalotaxine can inhibit pseudotyped virus entry and membrane fusion in a dose-dependent manner, with enhanced effectiveness upon elevated expression of TMPRSS2. Importantly, they displayed low cytotoxic effects on human normal cell lines. Our study suggested that 5'-de-O-methylharringtonine and cephalotaxine were of low toxicity and safety for humans as potential antagonists of SARS-CoV-2 and its variants, which deserve further validation in a biosafety level 3 facility.