Absent of drugs to treat entervirus infections, notably the coxsackievirus B3 virus (CVB3) which causes acute and chronic illnesses, the world remains in need of new antiviral drugs. The main objective of this work is the quantitative analysis of the structure-activity relationship (QSAR) of a series of phenyloxy propy isoxazoles derivatives against the CVB3 virus using two 2D approaches using the HQSAR method and 3D using the Topomer CoMFA and CoMFA and CoMSIA methods, followed by molecular docking analysis to validate established patterns and understand the mechanism of receptor-ligand interaction.  The results of the 2D / 3D QSAR models are quite satisfactory and give significant statistical results: R² = 0.953, Q² = 0.819, R²_ext = 0.750 for the HQSAR, R² = 0.980, Q² = 0.83, R²_ext = 0.749 for the CoMFA topomer, R² = 0.977, Q² = 0.748, R²_ext = 0.843 for CoMFA, R² = 0.962, Q² = 0.804, R²_ext = 0.953 for CoMSIA.  It can be noted that these four models exceeded the external validation criteria used with success and respected the limits of the criteria of Tropsha and Glorbaikh. Based on the results obtained from the four models, we proposed a candidate for each model as an inhibitory agent against CVB3. Docking analyzes and molecular simulation were performed to understand the mechanism of interactions of these four designed compounds within the receptor active site. small-sized electron donor groups molecular docking shows that the proposed compounds performed greater interactions than the more active compound in the database. However, the groups added for the molecules A1, A2, A3, A4 help to create additional interactions between these ligands and the residues to stabilize the conformation of the ligands at the level of the binding pocket. The stability and binding modes of compounds A1, A2 and the most active compound in the data set were evaluated by molecular dynamics simulations during a simulation time of 100 ns. It is shown that the interactions of the selected compounds are stable and fluctuate weakly in the complex. Free energy calculations based on the MM-GBSA method confirmed that the two designed compounds A1 and A2 were able to form bonds in the protein cavity. In addition, the ADMET study and the five-parameter Lipinski's rule prediction were estimated to ensure that the proposed candidates are viable drugs with synthetic accessibility. These results can be used for the discovery of new drugs and can solve the problem of resistance of the CVB3 virus