AbstractDuring the first years of COVID‐19 pandemic, X‐ray structures of the coronavirus drug targets were acquired at an unprecedented rate, giving hundreds of PDB depositions in less than a year. The main protease (Mpro) of severe acute respiratory syndrome‐related coronavirus 2 (SARS‐CoV‐2) is the primary validated target of direct‐acting antivirals. The selection of the optimal ensemble of structures of Mpro for the docking‐driven virtual screening campaign was thus non‐trivial and required a systematic and automated approach. Here we report a semi‐automated active site RMSD based procedure of ensemble selection from the SARS‐CoV‐2 Mpro crystallographic data and virtual screening of its inhibitors. The procedure was compared with other approaches to ensemble selection and validated with the help of hand‐picked and peer‐reviewed activity‐annotated libraries. Prospective virtual screening of non‐covalent Mpro inhibitors resulted in a new chemotype of thienopyrimidinone derivatives with experimentally confirmed enzyme inhibition.