Quantitative structure–activity relationship studies have been performed on twenty one β-carboline derivatives to investigate the structural requirements for antitumor activity. The best 2D-QSAR model (r2 = 0.802, F = 24.321, r2se = 0.325) indicated statistical significance and internal predictivity of the developed model shown by the value of cross validated squared correlation coefficient which was 0.724. A five-point pharmacophore hypothesis yielded ligand based pharmacophore 3D-QSAR models with good partial least-square (PLS) statistics results. The training set and test set correlation was characterized by PLS factors (r 2 = 0.842, SD = 0.306, F = 21.3, P = 4.27e -05, Q2 ext = 0.748, RMSE = 0.531, Pearson-R = 0.975). A docking study revealed the binding orientations of DNA intercalates at active site of amino acid residues. The results of 2D-QSAR and 3D-QSAR give detailed structural insights and at the same time highlight the important binding features of novel β-carboline derivatives as antitumor agents.