The structural features of retinoids and their relationship with biological targets and ADMET properties were explored using various computer programs. Due to their lipophilicity most investigated retinoids violate Lipinski ‘Rule 5’ which reflects in reduced oral absorption and bioavailability. The mechanism of retinoids action involves binding to nuclear receptors that act as transcription factors which was also confirmed by drug-likeness with nuclear receptor ligand (NRL dls). In addition to NRL dls, the enzyme inhibitor drug-likeness (EI dls) has been also computed by Molinspiration software (except for seletinoid G). For 1st generation of retinoids the following biological targets with high probability (100%) were predicted by Swiss Target Prediction software, i.e., retinoic acid receptors (RAR and RXR), mitogenactivated protein kinase 1, cellular retinoic acid binding proteins (CRABP) 1 and 2 and ornithine decarboxylase. However, these targets were predicted with lower probability (55%) for retinol in addition to retinol-binding plasma protein 1-176 (RBP4) (100%) and mitogenactivated kinase 8-11 and 14 (50%). Isotretinoine differs from other retinoids in 1st generation since zero probability was predicted for its interaction with RXR receptors, whereas its most likely targets are mitogen-activated kinase 1, 8 and 14 (100%). According to predicted target probabilities the most similar retinoids from 2nd to 4th generation to those in 1st generation are in the folowing order fenretinide > bexarotene> adarotene> pelretin, while for etretinate, motretinide, tamibarotene, mofarotene and seletinoide G were predicted zero probabilities for typical targets of 1st generation retinoids. Using ADMET PredictorTM 8.0 the following toxicity risks were predicted: ADMET risk from 1 (seletinoide G) to 9 (etretinate), CYP risk from 0 (seletinoid G) to 3 (retinol), TOX risk from 0 (tamibarotene and temarotene) to 3 (palovarotene) while the risk of mutagenic effect (TOX MUT risk 1) was predicted only for adapalene, tazarotene and seletinoid G. QSAR studies revealed significant correlations (R2 = 0.8067 - 0.9917) between several molecular descriptors (natoms and Mr) and topological indices (TI). Insignificant correlations were found with predicted drug-likeness scores (dls) except for less significant nON with ICM dls and NRL dls (R2 = 0.6989 and 0.6983) and also with ADMET toxicity parameters. These results refer to the significant differences in retinoid structures and possibly other mechanisms of action and side effects in the group of investigated retinoids, especially for seletinoid G.