Quantitative structure-retention relationship (QSRR) models with high predictive capability were constructed using published retention index (I) values for 39 iso-butoxycarbonyl derivatives on a DB-5 stationary phase and published values of retention times (t_r) for 26 t-butyldimethylsilyl phenol derivatives on an HP-5 stationary phase. Two QSRR models (I & II) were constructed on the basis of ten Dragon descriptors and retention time values for both derivatives using multiple linear regression (MLR). Model I (26 members) had the best statistical parameter fit (R = 0.988, SD = 1.568, F = 49.09, and SSE = 0.675). Five structural descriptors calculated on the basis of the number and position of chlorine substituents were used to derive two other models (III & IV) for t-butyldimethylsilyl derivatives on HP-5 and Rtx-200 stationary phases. The models based on structural descriptors exhibited more predictive capability than those obtained using Dragon and LSER descriptors.