The taxanes, paclitaxel and docetaxel, have favorable response rates in patients with breast, gynecologic, and lung cancers and have demonstrated activity against a variety of malignancies. In human trials, paclitaxel pharmacokinetics are nonlinear and are best fit by a three-compartment model with nonlinear distribution into the second compartment as well as nonlinear elimination. This finding is important for patients receiving paclitaxel at high doses or as a short infusion, as it results in disproportionately high peak concentrations and delayed elimination. The presence of nonlinear processes in docetaxel pharmacokinetics has not previously been examined. Therefore, plasma concentration data obtained from 53 patients receiving docetaxel at 55-115 mg/m2 over 1-24 h as part of phase I studies were modeled using the nonlinear three-compartment model found most suitable for paclitaxel and the results were compared with those obtained using the linear version. Docetaxel disposition was best described by the three-compartment nonlinear model in 28 of 53 data sets (53%). However, the difference in curve fit observed between the two models was modest (did not improve Akaike criteria) and unlikely to be of relevance. This study suggests that nonlinear processes in docetaxel pharmacokinetics may exist, but, unlike the case of paclitaxel, they are not likely to have a significant impact at the dose and administration schedule used in routine clinical practice (60-100 mg/m2 given over 1 h by infusion). The presence of nonlinear docetaxel pharmacokinetics at doses above 115 mg/m2 will have to be determined in case of further dose escalation.