Most patients with ovarian cancer currently are treated primarily with surgery and chemotherapy. Drug selection usually is not based on individualized in vitro sensitivity studies but on reported response rates of clinical trials. Attempts to include in vitro chemosensitivity testing into the management of ovarian cancer have been disappointing to clinicians. Tumor cells from fresh human ovarian cancer do not grow well under artificial in vitro growth conditions. The selection of cells that happen to proliferate in vitro (e.g., human tumor clonogenic assay) has resulted in low plating efficiencies (0.001-0.1% of plated cells). The vigorous mechanical and enzymatic tumor disaggregation, done to obtain a single-cell suspension, further reduces the number of cells that grow in vitro, resulting in low overall evaluability rates of 40-70% for the human tumor clonogenic assay. At the University of Miami, a new in vitro chemosensitivity assay was developed that detected the decrease in total tumor cell viability by measuring intracellular adenosine triphosphate as a function of in vitro drug response. Preliminary data on 31 tumor tissues from patients, which was evaluated with this method, showed a sensitivity of 92% and a specificity of 100%. Since these initial studies, data were gathered on more than 150 fresh gynecologic tumor specimens to evaluate single drugs and drug combinations at five concentrations (range, 10-500% of reported peak plasma concentrations). The evaluability rate for ovarian tumors was more than 90%. Some tumors showed almost complete cell kill at the lowest drug concentration; others had only a limited response at the highest level. Drug-response patterns also were variable for combined drug exposure. These findings underscore the heterogeneity of drug response in morphologically similar tumors and the importance of characterizing individual chemosensitivity profiles for patients before drug treatment.