To optimize the human tumor stem cell assay (HTSCA) for clinical and research purposes we have carried out in vitro pharmacology studies. Useful observations were made in four areas. (1) Drug assay design: The predictive accuracy of the HTSCA depends on the in vitro testing of drug concentrations of less than 10% of those which are pharmacologically achievable with standard in vivo drug doses. The use of unrealistically high in vitro concentrations can accurately predict clinical drug resistance, but is likely to yield high false-positive rates of clinical response prediction. (2) Drug scheduling: For certain schedule-dependent drugs, as well as those with a prolonged plasma half-life and those used according to a repeated daily schedule, prolonged in vitro exposure (rather than 1 h) may be needed to provide an adequate in vitro design. For an accurate prediction of sensitivity of tumor colony-forming units (TCFUs) to continuous drug contact in the agar, concentrations should be in the range of 1/300 that used for the standard 1-h exposure prior to plating. (3) Drug combinations: In preliminary studies of combination chemotherapy in vitro we commonly observed at least additive effects with low doses of cis-platinum plus either vinblastine or adriamycin. (4) Drug bioactivation: Rat liver microsomes or S-9 fraction were used to activate cyclophosphamide for in vitro effect, and satisfactory dose-response curves were observed for the inhibition of TCFUs. Such pharmacologic studies will be required for a wide variety of standard and new agents and will probably become a regular aspect of investigation of new anticancer drugs.