Abstarct: Herbaceous essential oils and related compounds have a broad application as for flavoring food, folk medicine and perfume and medical industries. In this project, the quantitative structure-property relationship studies was taken for retention indices of 77 and 119 essential oils in first and second parts, respectively. For modeling the retention indices of these compounds, two methods containing linear (multiple linear regression; MLR) and nonlinear (artificial neural network; ANN) were applied. The stepwise regression method was used for descxriptor selection. The models were validated using external test set, as well as leave one out and Y-Randomization techniques. Both linear and nonlinear methods have good prediction ability, although ANN model has more accurate results. The correlation coefficient (R) of the test set obtained by MLR and ANN models in the first part of this study were 0.975 and 0.982 respectively. Also in the second part, the correlation coefficient of the test set obtained by MLR and ANN models were 0.941 and 0.956 respectively.