Virtual screening of chemical databases is an emerging approach in drug discovery that uses computers to dock chemicals into the active site of a drug target to identify leads through evaluation of binding affinities of the chemicals. However, there are concerns about the validity and scope of the reported virtual screens due to lack of studies to show that randomly selected chemicals are not equally active and due to the fact that metalloproteins were rarely used as drug targets. We have performed a virtual screening of a chemical database to identify prototypic inhibitors of farnesyltransferase (FT) with zinc present in the active site. Among the 21 compounds identified by computers, four inhibited FT in vitro with IC(50) values in the range from 25 to 100 microM. The most potent inhibitor also inhibited FT in human lung cancer cells. In contrast, none of 21 randomly selected compounds have an IC(50) lower than 100 microM. The results demonstrate the validity of virtual screening and the feasibility of applications of this approach to metalloprotein drug targets, such as matrix metalloproteinases, farnesyltransferase, and HIV-1 integrase, for the treatments of cardiovascular diseases, cancers, and AIDS.