Several recent developments suggest that the human glyoxalase I (GLO I) is a potential target for antitumor drug development. In present study, a series of curcumin derivatives with high inhibitory activity against human GLO I were discovered. Inhibition constant (KO values of compounds 8, 9, 10, 11 and 13 to GLO I are 4.600 mu M, 2.600 mu M, 3.200 mu M, 3.600 mu M and 3.600 mu M, respectively. To elucidate the structural features of potent inhibitors, docking-based three-dimensional structure-activity relationship (3D-QSAR) analyses were performed. Satisfactory agreement between experiment and theory suggests that comparative molecular similarity index analysis (CoMSIA) modeling exhibit much better correlation and predictive power. The cross-validated q(2) value is 0.638 while no-validation r(2) value is 0.930. Integrated with docking-based 3D-QSAR CoMSIA modeling, molecular surface property (electrostatic and steric) mapping and molecular dynamics simulation, a set of receptor-ligand binding models and bio-affinity predictive models for rational design of more potent inhibitors of GLO I are established. (C) 2010 Elsevier Ltd. All rights reserved.