A combination between 3D-QSAR and molecular mechanics (MM)-docking study was used as a tool to detail and model the mechanism of action of 46 antifungal azoles. Two methods of alignment of the ligands were performed (i) alignment of the main skeleton without substituents and (ii) alignment of a defined substructure. The best model is characterized by q(2) with the values of 0.70 for yeastlike (yeast), 0.66 for filamentous fungi, and 0.70 for the selectivity against filamentous fungi. 3D-QSAR regression maps derived from six models were used to identify the regions responsible for the differences in the compounds activity against yeast and filamentous fungi. The binding energy of the important substructures (Local Binding Energy-LBE) and its standard deviation were calculated in order to demonstrate quantitatively the contribution of substituents reflecting the diversity of the antifungal activity. The comparisons of these results with the same regions of the contour maps indicated a good correspondence between the 3D-QSAR and MM (LBE) approaches allowing association between the maps and the participating residues in the active sites of P450DM of C. albicans and A. fumigatus. The pi-pi interactions of two or more aromatic groups of the ligands with Phe228 and Tyr132 prove to be most important for the differences in activity against C. albicans. In A. fumigatus there was a better occupation of the inner central I-spiral in the areas around the heme. For the activity against A. fumigatus the pi-pi interactions of aromatic groups of the compounds with Phe509, Phe228, and Tyr132 are significant for the activity.