Taxol, a natural product initially isolated from the stem bark of the western yew Taxus brevifolia, is undergoing phase II and III evaluation due to its reported activity against a variety of tumors. Previous studies have described correlations between plasma concentrations and toxicity when taxol is given (a) at lower doses, (b) for shorter infusion times, and (c) without granulocyte-colony-stimulating factor. Because the 24-h infusion schedule is most commonly used in current clinical trials, we attempted to correlate steady-state plasma concentrations of taxol achieved with a 24-h continuous i.v. infusion with toxicities and responses. Plasma samples from 48 refractory ovarian cancer patients were obtained 1-2 h prior to the end of the first taxol infusion. Taxol concentrations were measured by high-performance liquid chromatography (HPLC). Interpatient variation of taxol plasma concentrations was small (mean +/- SD, 0.85 +/- 0.21 microM. Total taxol body clearance was 256 +/- 72 ml min-1 m-2 (mean +/- SD). Taxol plasma protein binding was 88.4% +/- 1.3% (mean +/- SD, n = 9). Grade 3-4 hematologic toxicity, mainly leukopenia, occurred in 92% of the patients. The leukopenia was transient and did not warrant a reduction in the dose of taxol. Grade 3-4 nonhematologic toxicity occurred in 8% of the patients. No severe hypersensitivity reaction or grade 3-4 neurotoxicity was observed. Correlations of plasma concentrations and toxicities were not feasible due to the high frequency of hematologic effects and the low frequency of nonhematologic toxicity. The low degree of interpatient variation in plasma concentrations hindered the development of correlations with response.