The present study aimed to conduct a quantitative analysis of structure-retention relationship, using artificial neural network (ANN) and multiple linear regression (MLR) models for determining the retention time of some organic chemicals on gas chromatography in a modified carbon nanotube (CNT) capillary column, prepared by CVD method. The dataset consisted of 37 compounds. The data were categorized into training, external, and internal datasets, including 28, four, and five molecules, respectively. Qmean, XMOD, TE2, H050, qnmax, and LBW were the optimal descriptors in the model, encoding different molecules involved in steric and electronic interactions. In the optimal model, Rtrain, Rinternal, and Rexternal were 0.968, 0.965, and 0.953 for MLR and 0.984, 0.983, and 0.964 for ANN, respectively. Using the leave-many-out cross validation method, as well as y-scrambling test, the model's reliability was determined (Q2 = 0.926 and 0.969; RMSE = 2.363 and 1.513 for MLR and ANN models, respectively). Our findings revealed the efficacy of the developed QSRR models in predicting the retention time of different chemicals, based on their measured molecular descriptors.