Our aim was to develop a population pharmacokinetic model of ultrafilterable oxaliplatin in metastatic cancer patients. Oxaliplatin was administered by 2- or 4-h infusions, 50, 65, 75, 85, 100 or 130 mg/m2 to 56 patients. Blood samples were collected over 28 h. Plasma concentrations of ultrafilterable oxaliplatin were determined by flameless atomic absorption spectrophotometry. Population pharmacokinetic analysis was performed using a non-linear mixed-effects modeling method. Ultrafilterable oxaliplatin concentration-time profiles showed a secondary peak or a shoulder aspect post-infusion, attributed to the existence of an enterohepatic recirculation (EHR). They were best described by a two-compartment model incorporating an EHR component. Plasma clearance (CL) was related positively to body weight (BW) and negatively to serum creatinine (SCr), and was greater in male patients than in female patients. This covariate modeling resulted in a decrease in the interindividual variability for CL from 104 to 62%. The central distribution volume (V1) and inter-compartmental clearance (Q) were related to BW. Typical population estimates of CL, central distribution volume (V1), input rate constant into gallbladder (k1B) and lag time for drug reabsorption (TLAG) were 14.1 or 8.5 l/h (male or female patients), 24.9 l, 1.8 h-1 and 2.0 h, respectively. The final pharmacokinetic model was validated using 200 bootstrap samples of the original data. We conclude that a two-compartment with EHR model adequately described ultrafilterable oxaliplatin pharmacokinetics, explaining a secondary transient increase in concentration. This study identified combined-covariate-effects ultrafilterable oxaliplatin clearance, supporting dose adjustment of oxaliplatin based on BW, gender and corrected for SCr level, if drug exposure is thought to be related to therapeutic or toxic issues.