For the COVID-19 pandemic caused by SARS-CoV-2, there are currently no effective drugs or vaccines to treat this coronavirus infection. In this study, we focus on the main protease enzyme of SARS-CoV-2, 3CL pro , which is critical for viral replication. We employ explicit solvent molecular dynamics simulations of about 150 compounds docked into 3CL pro ’s binding site and that had emerged as good main protease ligands from our previous in silico screening of over 1.2 million compounds. By incoporating protein dynamics and applying a range of structural descriptors, such as the ability to form specific contacts with the catalytic dyad residues of 3CL pro and the structural fluctuations of the ligands in the binding site, we are able to further refine our compound selection. Fourteen compounds including estradiol shown to be the most promising based on our calculations were procured and screened against recombinant 3CL pro in a fluorescence assay. Eight of these compounds have significant activity in inhibiting the SARS-CoV-2 main protease. Among these are corilagin, a gallotannin, and lurasidone, an antipsychotic drug, which emerged as the most promising natural product and drug, respectively, and might thus be candidates for drug repurposing for the treatment of COVID-19. In addition, we also tested the inhibitory activity of testosterone, and our results reveal testosterone as possessing moderate inhibitory potency against the 3CL pro enzyme, which may thus provide an explanation why older men are more severely affected by COVID-19.