A molecular modeling study was carried out to develop a predictive model for combretastatin-like analogues populating the colchicine-binding site of beta-tubulin. A series of compounds built around a framework including two aromatic groups linked by various moieties such as alkenes (stilbenes), enones (chalcones), or ethers was selected for the study. The 5D-QSAR model was developed stepwise. First a model was generated for the chalcone series (19 compounds, 71 conformations), then for the stilbene series (18 compounds, 59 conformations), and finally for the combined dataset (47 ligands, 160 conformers). Although the models for the chalcone and stilbene series appeared slightly different when represented by QSAR colored surfaces, the combined model seems to reconcile the differences without compromise and represents a highly predictive model for compounds that bind to the colchicine-binding site of tubulin.