The Quantitative Structure Retention Relationship (QSRR) modeling techniques are employed for prediction of retention behavior of chiral secondary alkylaromatic and alkylheterocyclic alcohols, derivatives of 1-phenylethanol, separated on Chiracel OB-H column. Genetic algorithms and neural networks are used to obtain models predicting Retention Order Index (ROI) (R(2) - 0.99), selectivity ROI log alpha (R(2) - 0.93) as well as retention factors (log k) for two types of mobile phases (90/10 and 85/15 n-hexane/isopropanol--R(2) - 0.97 and 0.95). Additionally, a model that predicts log k for both mobile phase in function of i-PrOH concentration is developed (R(2) - 0.97). HOMO energy turns out to be the most important parameter in description of log k while mixed steric-electrostatic interactions with chiral OH group and furan ring are responsible for the chiral recognition. The models are used to assess the stereoselectivity of ethylbenzene dehydrogenase (EBDH), which catalyzes stereospecific syntheses of the investigated compounds. The high stereoselectivity of the enzyme is confirmed but reversion of EBDH enantioselectivity is predicted to take place in the biosynthesis of 1-[1,1'-biphenyl]-4-ylethanol.