Hsp90 (Heat shock protein 90) is an important therapeutic target for the treatment of cancer. To identify important chemical features for Hsp90 inhibitory activity, a 3D-QSAR pharmacophore model was developed using a set of 61 inhibitors (a training set of 31 and a test set of 30 compounds) belonging to a series of 2-amino-6-halopurine and 7'-substituted benzothiazolothio- and pyridinothiazolothio-purines. The best HypoGen model consisted of five pharmacophoric features: one hydrogen bond acceptor (HBA), one hydrogen bond donor (HBD) and three hydrophobic (HY) groups. It showed a high correlation coefficient (r = 0.943) and low root mean square deviation (RMSD = 0.751). This model was validated against 30 known Hsp90 inhibitors, where it showed a high predictive value for R(2)pred = 0.805], thus confirming that HY, HBA and HBD features are essential for Hsp90 inhibition. The validated pharmacophore model (Hypo-1) was used as a 3D query for virtual screening to retrieve potential inhibitors from the Maybridge and National Cancer Institute (NCI) databases. The hit compounds were subsequently subjected to molecular docking studies and, finally, five hits were prioritized as potential leads based on GoldScore function.