OBJECTIVE Our aim was to evaluate the pharmacokinetics and pharmacodynamics of paclitaxel (Taxol; Bristol-Myers Squibb Co, Princeton, NJ) in children, and to determine whether paclitaxel exhibited saturable pharmacokinetics. METHODS We evaluated the pharmacokinetics and pharmacodynamics of paclitaxel (200 to 420 mg/m2) administered as a 24-hour intravenous (i.v.) infusion in a phase 1 study of 30 pediatric patients (age, 2.3 to 22.8 years) with refractory solid tumors. Fourteen serial blood samples were obtained during and up to 48 hours after the infusion, and paclitaxel concentrations were measured by a high-performance liquid chromatography-UV (HPLC-UV) method. Four pharmacokinetic models were compared for their ability to describe the patients' data. RESULTS Paclitaxel disposition was not consistent with a first-order, two-compartment pharmacokinetic model. Rather, the majority of data sets were best described by a two-compartment model that incorporated both saturable tissue distribution and saturable elimination; a smaller number of patient data sets were best described by models that incorporated either saturable distribution or saturable elimination. Clearance was dose-dependent, with a median clearance at the lower dosages (< 400 mg/m2) of 161 mL/min/m2, and at the highest dosages (> 400 mg/m2) of 123 mL/min/m2 (P = .044). The duration that paclitaxel plasma concentrations exceeded 0.1 mumol/L was highly variable (range, 26 to 71 hours). There was a trend toward higher median area under the concentration-versus-time curve (AUC) in those children with musculoskeletal (72 mumol/L.h; P = .054) or neurologic toxicity (54 mumol/L.h; P = .062) versus those without toxicity (30 mumol/L.h). Toxicity was not significantly correlated with dosage. CONCLUSIONS We conclude that paclitaxel distribution and elimination are saturable, and that estimates of paclitaxel systemic exposure correlate better with toxicity than does dosage.