BRD4 plays an indispensable role in cell cycle regulation, affecting processes such as cell proliferation, apoptosis and transcription. In this article, a three-dimensional quantitative conformational relationship (3D-QSAR) was used to investigate the molecular simulations related to 45 tetrahydrooxazole BRD4/BD2 selective inhibitors. 3D-QSAR models were developed based on two different analytical methods, COMFA and COMSIA. All CoMSIA field combinations were compared together and the best CoMSIA model was selected based on external validation and model prediction results. Both CoMSIA (S + E + H + D (q 2 = 0.808, r 2 = 0.99)) and CoMFA models (q 2 = 0.779;,r 2 = 0.962) predicted the internal and external well.Twenty small molecules with higher inhibitory activity than the template compound were designed by 3D-QSAR model prediction and quantum chemistry analysis, screened by molecular docking and ADMET methods, and five novel compounds were found to have better predicted activity and binding affinity. Molecular dynamics simulations showed that PRO70, PRO371, TYR428, ASN429 and VAL435 are important residues that can interact with the ligands. Moreover, further structural optimization of the more active candidate compounds will provide an essential theoretical basis of the synthesis and design of novel BRD4/BD2 inhibitors.