Abstarct: Nowadays quantitative structure-property relationship (QSPR) is used for studying the dependence of various physico-chemical properties of compounds to molecular structure. In the first section of this project, Multiple linear regression (MLR) as a linear method and artificial neural network (ANN) as nonlinear method used to predicting the Henry constant of 333 organic compounds. The data set was divided into a training (263 compounds), test (40 compounds) and an external validation (30 compounds) sets. A three-laxyered feed forward ANN with back-propagation of error was generated using four molecular descxriptors appearing in the MLR model. The statistical characteristics provided by multiple linear model (R2 = 0.942; MSE = 0.109 for test set) indicated predictive ability, while the predictive ability of ANN model is somewhat superior (R2 = 0.982; MSE = 0.039 for test set). In the second part of this project, genetic algorithm – multiple linear regression (GA-MLR) and genetic algorithm–artificial neural network (GA-ANN) methods were used to generate QSPR models between the descxriptors and refractive index of 144 polymers. The R2e, R2p, Mor15e and J3D descxriptors selected by genetic algorithm for constructing of linear and nonlinear models to predicting the refractive index. The root mean square error of the training, test, and validation sets for the ANN model are 0.018, 0.023, 0.018 and also determination coefficients (R2) of 0.929, 0.881, and 0.931 respectively.