To discover new anticancer drugs the U.S. National Cancer Institute (NCI) currently screens roughly 10,000 compounds annually against a panel of 60 human tumor cell lines. Drug activity is determined using a 2-day growth inhibition assay with the cells maintained in contact with the drugs for the 2-day period. Analyses of the more than 60,000 compounds screened to date against this panel of cell lines have revealed surprisingly detailed correlations between drug activities and the molecular genotypes and phenotypes of the cells. For example, most of the drugs, particularly those active in the clinic, are most active against p53 wild-type cells. This commentary points out, however, that these patterns of activity, and by implication the compounds identified as active, may not be relevant to tumor response in vivo. This is because tumor response is determined by tumor cell kill, and there is no reason to suppose that drug-induced short-term growth inhibition should correlate with drug-induced cell kill, usually measured by the ability of the cell to divide indefinitely to form a colony. We have tested a subset of the panel of NCI cell lines for cell kill by clonogenic assay using cisplatin and mitomycin C and have found only a weak and nonsignificant correlation between cell kill and activity determined by the NCI using the growth inhibition assay. It is concluded that a more extensive comparison of the 2-day growth inhibition and clonogenic assays should be made, as well as a test of which is more relevant to the response of human tumor xenografts.