Inhibition of transmembrane serine protease 2 (TMPRSS2) is expected to block the spike protein-mediated fusion of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Nafamostat, a potent TMPRSS2 inhibitor as well as a candidate for anti-SARS-CoV-2 drug, possesses the same acyl substructure as camostat, but is known to have a greater antiviral effect. A unique aspect of the molecular binding of nafamostat has been recently reported to be the formation of a covalent bond between its acyl substructure and Ser441 in TMPRSS2. In this study, we investigated crucial elements that cause the difference in anti-SARS-CoV-2 activity of nafamostat and camostat. In silico analysis showed that Asp435 significantly contributes to the binding of nafamostat and camostat to TMPRSS2, while Glu299 interacts strongly only with nafamostat. The estimated binding affinity for each compound with TMPRSS2 was actually consistent with the higher activity of nafamostat; however, the evaluation of the newly synthesized nafamostat derivatives revealed that the predicted binding affinity did not correlate with their anti-SARS-CoV-2 activity measured by the cytopathic effect (CPE) inhibition assay. It was further shown that the substitution of the ester bond with amide bond in nafamostat resulted in significantly weakened anti-SARS-CoV-2 activity. These results strongly indicate that the ease of covalent bond formation with Ser441 in TMPRSS2 possibly plays a major role in the anti-SARS-CoV-2 effect of nafamostat and its derivatives.