Abstarct: Quantitative structure–property relationship (QSPR) is an efficient method to predict the properties of various compounds baxsed on their molecular structure. In this study, QSPR models were developed to predict the retention time of a set of highly volatile organic compounds (VVOCs) using molecular descxriptors. For this purpose, a comprehensive set of descxriptors was calculated and then reduced using stepwise regression (SR), genetic algorithm (GA), random forest (RF), and glowworm algorithm (FF) as variable selection methods. Artificial neural networks (ANN) was used to model the potential nonlinear relationships between molecular descxriptors and retention time. The performance of the models was evaluated baxsed on various statistical criteria, including coefficient of determination and error-baxsed indices, for each external test set. The results showed that GA–ANN models have a relative advantage compared to other models, indicating the predictive power of this model, although other models also provided acceptable results. However, the SR–ANN model was able to achieve this level of performance using a smaller number of descxriptors, resulting in a simpler, more computationally efficient model. The findings of this study confirm the effectiveness of QSPR models baxsed on artificial neural networks and show that stepwise regression can be an effective strategy for selecting descxriptors without reducing the predictive ability of the model. The proposed approach provides a reliable tool for predicting the retention time of VVOC compounds and can significantly reduce the need for extensive and costly experimental measurements.