Abstarct: In the first section of this study, quantitative structure – activity relationship (QSAR) models were constructed for predicting the anti-HIV activity of some pyridinone derivatives. The group contribution and the stepwise multiple linear regression methods were used for variable selection. Then the best descxriptors were applied as input for QSAR modeling using random forest (RF) and artificial neural network (ANN) techniques. In modelling methods baxsed on random forest the data set was divided into training set and test set including 43 and 10 compounds, respectively. Effective parameters of RF method including Ntree and Mtry were optimized using the training set. For ANN modeling the data set was divided into training set, validation set and test set including 37, 8 and 8 compounds, respectively. ANN model was optimized too and the best model was selected. The validation study of the RF and ANN models was performed by prediction of the activity of the test set compounds and the correlation coefficients for RF and ANN model are 0.929 and 0.935 respectively. In the second section, RF and ANN models were constructed and the ability of prediction of anti-HIV activities for some of thiazole derivatives were investigated. The data set was divided into training and test set containing 18 and 6 chemical, respectively. The stepwaise regression method was used to select the most important descxriptors. The validation study of the RF and ANN models was performed using test set compounds. The correlation coefficients obtained for test set were 0.982 and 0.896 for RF and ANN respectively. The results obtained showed proper prediction power of the proposed models.