The damage to public health posed by the COVID-19 epidemic remains non-negligible, and research into highly effective antiviral drugs is key to addressing the epidemic. The main protease (Mpro) is critical for SARS-CoV-2 replication and acts as an effective target for drug development. To date, many flavonoid compounds have been reported to exhibit inhibition of Mpro, and flavonoid structures are informative for the development of new inhibitors. In the present study, we first evaluated the ability of 35 flavonoid compounds to bind Mpro and found that the hydrophobic interaction generated between the benzene ring and the residues may be the main source of binding force. The in vitro inhibitory activity of these compounds was tested by the FRET method, and most of them showed significant inhibition of Mpro. Preliminary analysis revealed that the presence of the glycosidic group was detrimental to the activity of the compounds, in addition the position of the hydroxyl substituent had an effect on the activity of the compounds. Reliable and predictive CoMFA (q2 = 0.541, r2 = 0.998, SEE = 0.039) and CoMISA (q2 = 0.554, r2 = 0.999, SEE = 0.034) models were developed, and the compounds were analyzed by 3D-QSAR with modeled steric, electrostatic, hydrogen bond donor, and hydrophobic fields. A series of compounds were synthesized and characterized, which showed similar experimental and predicted activities, with the activity of 15d (IC50 = 0.74 ± 0.04 uM) exceeding that of the template molecule 15 (0.98 ± 0.13 uM). Subsequently, molecular docking revealed the binding mode of 15d to Mpro and verified the predictions of steric, hydrogen bond donor, and hydrophobic field. In conclusion, these findings provide ideas for subsequent structural studies of flavonoids that could be used in the development of flavonoid-type SARS-CoV-2 Mpro inhibitors.