Neutropenia is the main dose-limiting toxicity occurring in docetaxel treatment. The objective of this study was to identify pharmacodynamic (PD) factors responsible for the neutropaenia caused by docetaxel. Data were obtained from 92 patients treated with docetaxel as a monochemotherapy in two different treatment centres. A semiphysiological population pharmacokinetic-pharmacodynamic (PK/PD) model was applied to describe the time course of neutrophils and the neutropaenic effect of docetaxel. The plasma docetaxel concentration was assumed to inhibit the proliferation of neutrophil precursors through a linear model: Drug effect=Slope x Conc. Slope corresponds to the patients' sensitivity to the neutropaenic effect of docetaxel. Covariate analysis was performed by testing the relationship between the patients' characteristics and Slope using the program NONMEM. The neutropaenic effect of docetaxel showed a high interindividual variability. Three significant PD covariates were identified: serum alpha1-acid glycoprotein levels (AAG), level of chemotherapy pretreatment, and treatment centre. Extensive pretreatment was associated with an increase in Slope values meaning a higher haematotoxicity. An increase in AAG was associated with a decrease of both Slope and docetaxel plasma clearance. Patients treated in one centre had both higher Slope and docetaxel clearance. The centre effect (most likely due to a bias in the PK part of the study between the two centres) reveals the robustness of the PK/PD model. Individual dosing of docetaxel should be based on previous chemotherapy but not on the AAG level since it has a similar influence on PD and PK docetaxel parameters. This methodology should be applied to further investigate elderly patients and to identify more precisely the characteristics of previous chemotherapy that contribute to the cumulative myelotoxicity.