Paclitaxel shows a broad clinical activity in ovarian, breast and non-small cell lung cancers. However, controversy remains about the respective effects of doses and schedules in paclitaxel cytotoxicity. This study was conducted to compare the cytotoxic activity of short-term (1 h) versus prolonged exposure (14 days) to paclitaxel in human cancer cells. A soft-agar cloning system assay was used to determine the in vitro effects of 0.025-25.0 microg/ml paclitaxel against cancer cells taken directly from patients. A decrease in tumor colony formation resulting from drug exposure was considered an in vitro response if survival of colonies was up to 50% of that in positive controls. Among 11 evaluable patients' biopsies, both short- and long-term exposure to paclitaxel had significant concentration-dependent effects on the growth inhibition of human cancer cells. With the 1 h exposure schedule, in vitro responses were observed in 9,18 and 64% of evaluable tumor specimens at final concentrations of 0.25, 2.5 and 25.0 microg/ml, respectively. With the prolonged exposure schedule, concentrations of 0.25, 2.5 and 25.0 microg/ml induced 27, 45 and 91 in vitro response rates, respectively. In those patients' biopsies prolonged exposure to paclitaxel induced significantly more in vitro tumor responses than 1 h administration (p < 0.01). Similar trends were observed in ovarian, breast and non-small cell lung cancers. Our data indicate that the duration of exposure to paclitaxel is an important factor in paclitaxel cytotoxicity in human tumors and suggest that long-term exposure may improve the antitumor activity of paclitaxel.