Objectives: To develop a population pharmacokinetic (PK) model for simultaneous analysis of oral and intravenous data, to compare the variability between the two routes of administration of vinorelbine, to search for the main patient characteristics that explain this variability, and to estimate the mean population bioavailability of oral vinorelbine., Patients and Methods: A PK model was developed from 175 phase I/II patients (419 courses) treated by intravenous (20-45 mg/m2) and/or oral (60-100 mg/m2) vinorelbine given as monotherapy. Oral and intravenous PK data were simultaneously fitted using the NONMEM program, allowing the estimation of oral PK parameters such as the bioavailability factor in patients who received only the oral formulation. Covariates included demographic characteristics, biological markers, hematological parameters, liver metastases, early vomiting, and food intake. The population covariate model was developed from rich sampling data ( n=187 phase I courses) and then assessed from sparse sampling data ( n=232 phase II courses)., Results: A three-compartment model best described the combined oral/intravenous blood concentration-time data. The mean absolute bioavailability was 36%, with moderate interindividual (CV=20%) and intraindividual (CV=19%) variability. Bayesian clearance was accurately estimated in 180 of 187 patients. The clearance of oral and intravenous vinorelbine showed comparable variability at usual doses (25-30 mg/m2 intravenous; CV=26%; 60-80 mg/m2 oral, CV=33%) and was moderately increased when including maximum tolerated doses (20-45 mg/m2 intravenous, CV=27%; 60-100 mg/m2 oral, CV=36%). Several relevant covariate relationships influencing the total body clearance of vinorelbine were independent of the route of administration: body surface area (proportional relationship), platelet count above 400 x 10(9)/l (negative correlation), creatinine clearance (positive correlation), and elevated transaminases (negative correlation). Food intake induced a lag time in the absorption of oral vinorelbine. A weak and poorly estimated relationship was observed between elevated alkaline phosphatase levels and bioavailability, although hepatic markers such as GGT, LDH, total protein, and liver metastases and age had no effect on vinorelbine pharmacokinetics., Conclusions: By means of the simultaneous analysis of oral and intravenous data the bioavailability (F=36%) and its associated variability were estimated. At usual doses similar levels of variability were observed between oral and intravenous routes. As a result of the identification of covariates from phase I data and their confirmation from phase II data further explorations based on limited sampling strategies are now possible. The use of a simultaneous oral/intravenous model allows a better characterization of the PK profile of vinorelbine after administration by either vascular or oral route