25 saturated alcohols compounds molecular structure were optimized by chemoffice2004. 1897 molecular structural parameters were obtained by Dragon6.0. Two molecular descriptors (Chi_H2 and REIG) were selected using Minitab15 by stepwise regression method. Quantitative structure-retention relationships (QSRR) on single column were established by multiple linear regression with the correlation coefficient R＞0.98. The sum of the first five McReynolds phase constants combing Chi_H2 and REIG molecular descriptors were used to establish the cross-column QSRR. Cross-column retention prediction was evaluated on an internal validation set consisting of 5 alcohols collected with each of the six columns, sequentially excluded from calibration. The statistical analysis showed that the QSRR models have high internal stability and good predictive ability.