Purpose: To evaluate the population pharmacokinetics of pemetrexed disodium in cancer patients enrolled in four different open-label, multicenter, nonrandomized phase II studies. Methods: Pemetrexed disodium was administered as a 10-min intravenous infusion (600 mg/m(2)) every 21 days. A total of four blood samples were to be collected each cycle per patient (n = 103 patients) during cycles 1 and 3. Plasma concentration-time data were analyzed by nonlinear mixed-effect modeling using NONMEM to estimate pemetrexed disodium pharmacokinetic parameters (mean, and between- and within-patient variability) as well as relationships between the pharmacokinetic parameters and various patient-specific factors (demographic and physiologic data). Results/Conclusions: The pharmacokinetics of pemetrexed disodium were best characterized by a two-compartment model with initial distribution and terminal elimination half-lives of 0.63 h and 2.73 h, respectively. The typical value of systemic clearance (CL) in liters per hour included a relationship to creatinine clearance (CrCL) with a slope of 0.0292. Typical values of central volume (V-c), distributional CL (Q), and peripheral volume (V-p) were 11.3 l, 3.21 l/h, and 5.20 l, respectively. Between-patient variability was 19.6%, 15.6%, and 21.7% for CL, V-c, and V-p, respectively. A combined additive/proportional error model was used to describe residual variability, with a coefficient of variation of 23.7% for the proportional component and a standard deviation of 0.0410 mu g/ml for the additive component. Significant patient-specific factors on CL were calculated CrCL, body weight, and to a lesser extent alanine transaminase and folate deficiency. Gender and body weight were significant factors on V-c while both body surface area and albumin were significant factors on V-p. In conclusion, population pharmacokinetic modeling revealed relationships between pharmacokinetic parameters and various patient specific factors.