Cyclin-dependent kinases (CDKs) have been recognized as key regulators of cell cycle progression. Alteration and deregulation of CDK activity are pathogenic hallmarks of neoplasia. Therefore, inhibitors or modulators would be of interest to explore as novel therapeutic agents in cancer, as well as other hyperproliferative disorders. Flavopiridol is a semisynthetic flavonoid that emerged from an empirical screening program as a potent antiproliferative agent that mechanistic studies demonstrated to directly inhibit CDKs 1, 2, and 4 as a competitive ATP site antagonist. Initial clinical trials have shown that concentrations that inhibit cell proliferation and CDK activity in vitro can be safely achieved in humans, and additional clinical trials will establish its clinical potential. To address the need for additional chemotypes that may serve as lead structures for drugs that would not have the toxicities associated with flavopiridol, compounds with a similar pattern of cell growth inhibitory activity in the National Cancer Institute's in vitro anticancer drug screen have been recognized by the computer-assisted pattern recognition algorithm COMPARE and then screened for anti-CDK activity in a biochemical screen. The benzodiazepine derivative NSC 664704 (7,12-dihydro-indolo[3,2-d]benzazepin-6(5H)-one) was revealed by that approach as a moderately potent (IC50 0.4 microM) inhibitor of CDK2, which in initial experiments shows evidence of causing cell cycle redistribution in living cells. NSC 664704 is, therefore, a candidate for further structural optimization, guided in part by understanding of the ATP-binding site in CDK2. This approach represents one way of combining empirical screening information with structure-based design to derive novel candidate therapeutic agents directed against an important cellular target.