The critical relationship between quantitative structure-property relationships (QSPRs) and the quality of the experimental data is discussed by using multiple linear regression (MLR) analysis and describe the activities and influences of the vibrational frequency ( stretching & bending ) for groups C=C and =CH2 to predicated intrinsic viscosity by semiemperical molecular orbital methods AM1&PM3. Three models with R2 ranges from 0.830-0.991by using MLR based on descriptors calculated from PM3 semiemperircal. A highly significant two-parameter correlation (R2 = 0.991, F = 57.685 and low S =10.541 ), employs just two molecular parameters C=C str and =CH2 bending was eq 3., these parameters which indicate play an important role in effect on intrinsic viscosity of polymer. The capability to predict by these techniques for intrinsic viscosity, set a successful example for applying a similar approach in building QSPR models for intrinsic viscosity and that could potentially offer a new opportunity in the design of novel properties of polymers or extended to other polymer composite.
