21 randomly selected drugs and excipients of different molecular structures (nipagin M, nipagin P, theophyline, caffeine, triclosan, phenylbutasone, vitamin k3, indomethacin, benzophenone-4, lormetazepam, elenium, naproxen, ibuprofen, bromazepam, aspirin, medazepam, spironolactone, cortisone acetate, olanzapine, chloramfenicol, sumatriptan) were subjected to HPLC chromatography on two different stationary phases: RP-18 and RP-18Ar using the 50:50 (v/v) binary mixture of pH 7.4 phosphate-buffered saline – acetonitrile as a mobile phase. Skin permeability coefficients log K_p of studied compounds were predicted in silico using Pott’s model. The relationships between the chromatographic retention factors log k of compounds listed above obtained for both stationary phases and their predicted skin permeability were investigated. It was concluded that C-18 HPLC retention parameters are not suitable as sole predictors of skin permeability. Multivariate linear (MVL) models of log K_p obtained for the studied group of compounds account for up to 73% of total variability. Calculated and chromatographic descriptors in MVL models were selected in the following order: TPSA (topological polar surface area); #FRB (total count of freely rotatable bonds); log k_RP-18Ar (chromatographic retention factor); F_Csp3 (fraction of sp³ carbon atoms). Chromatographic retention factors on the RP-18Ar stationary phase are more useful than those obtained on RP-18.