The QSAR model for the prediction of cytotoxicity of Ionic liquids (ILs) was developed using a data set of 1195 compounds. The Artificial Neural Networks learning technique was used. The predictive ability of the models was tested by means of cross-validation; the q(2) value was 0.76 for the regression model. The prediction for the external evaluation set afforded high predictive power (q(2)=0.75 for 239 compounds). The developed QSAR models evaluated the anticancer activity of a small set of virtual compounds and 6 compounds were selected for synthesis and biological testing. It was found that imidazolium and pyridinium ILs with C-12 and C-10 alkyl chain length exhibited significant cytotoxicity, particularly, compounds 3 and 6 were identified as the most potent anticancer agents with IC50 values 0.18 mu M and 5.75 mu M against Hep-2 cell line and different acute toxicity levels to cladoceran Daphnia magna. Molecular docking showed that the high cytotoxic activity of imidazolium and pyridinium ILs with C-12 alkyl chain length may be associated with specific DNA binding in the region of CG nucleotides.