The purpose of present study was to develop a population pharmacokinetic model of high dose methotrexate (HD-MTX) infusion in patients with lymphoid malignancy, to investigate the biological and clinical covariates related to the drug distribution and elimination. It is also the purpose to propose a limited sampling strategy (LSS) for the estimation of the time above the threshold (0.2 micromol.L(-1)). A total 82 patients with lymphoid malignancy were involved in the study. A pharmacokinetic model was developed using nonlinear mixed-effect model. The influence of demographic characteristics, biological factors, and concurrent administration were investigated. The final predictive performance was validated by bootstrap and cross-validation. Bayesian estimation was evaluated. The pharmacokinetics of HD-MTX was described by a two-compartment model. The pharmacokinetic parameters and the inter-individual variability were as follows: the clearance CL, 7.45 L.h(-1) (inter-individual variability 50.6%), the volume of the central and peripheral compartment V(1), 25.9 L (22.5%), V(2), 9.23 L (97.8%), respectively, and the intercompartmental clearance Q, 0.333 L.h(-1) (70.4%). The influence of serum creatinine on CL and weight on V(1) was retained in the final model. The protocol involved one sampling time at 44 h after the start of the infusion, allowing one to predict the time at which the MTX concentration reached the expected threshold (0.2 micromol.L(-1)). Serum creatinine and weight showed significant influence on methotrexate CL and V(1), respectively. Furthermore, a Bayesian estimation based on the covariates and 44 h sample was developed, allowing prediction of the individual methotrexate pharmacokinetic parameters and the time to 0.2 micromol.L(-1).