The selection of property relevant conformers is among the crucial choices in the development of a 3D-QSPR model. A novel methodology for quasi-multi-way PLS mining of the property (S. aureus FabI inhibition activity) relevant conformers is reported in the highly predictive 3D-QSPR model development. The physicochemical (PC) properties-based QSPR model showed nonvalidated r2 (NV-r2) of 0.995, cross-validated leave-one-out ([image omitted]) of 0.936, and predictive r2 on six test compounds (pred r2) of 0.963. The molecular field analysis-based QSPR model developed, using the PC-based model selected bioactive conformers, exhibited NV-r2 of 0.979, [image omitted] of 0.619, and pred r2 on seven test compounds of 0.996. The novel method can be extended for development of any 3D-QSPR model to accurately predict any 3-D-related properties.