Four lipophilicity descriptors (RM0, b, C0, PC1) for twelve coumarine derivatives were determined by reversed-phase thin-layer chromatography in order to analyze which descriptor best describes the lipophilicity of coumarines investigated. Moreover, possible chemical toxicity of coumarins, expressed as the probability of a compound to cause organ-specific health effects, was calculated using ACD/Tox Suite program. The quantitative relationships between toxicity and molecular descriptors, including experimentally determined lipophilicity descriptors obtained in current study, were investigated using partial least square regression. The best models were obtained for kidney and liver health effects. Quantitative structure-toxicity relationship models revealed the importance of electric polarization descriptors, size descriptors and lipophilicity descriptors. Obtained models were used for the selection of the structural features of the compounds that are significantly affecting their absorption, distribution, metabolism, excretion, and toxicity. [Acknowledgements. This work has been supported by the Ministry of Education and Science of Serbia, Grant 172017.