Kováts and Lee retention indices were predicted by the help of experimental and calculated boiling points and also by linear solvation energy relationship (LSER) models. The developed approaches should be applied as classifiers in non-target analysis of environmental samples to identify contaminants. To demonstrate the application as a classifier, an example of 14 isomers with empirical formula C₁₂H₁₀O₂ was selected. Furthermore, seven compounds with different molecular composition were used to illustrate the application in non-target analysis, where progressive candidate exclusion is performed. The models help to reduce the number of potential candidates, and seem to be a useful addition to already existing classifiers.