Tacrolimus is an immunosuppressant commonly used in the prevention of graft-versus-host disease (GVHD) following allogeneic HCT. Unfortunately, the use of tacrolimus is associated with variable immunosuppression and toxicity. The purpose of this study was to describe tacrolimus population pharmacokinetic parameters, to identify relationships between clinical covariates and pharmacokinetic estimates, and to develop a model to predict tacrolimus clearance in HCT patients. Steady-state whole blood tacrolimus concentrations (n = 1625) obtained during intravenous and oral therapy were analyzed in 122 patients. Population clearance (CL) was 5.22 l/h and bioavailability (F) was 0.28. The influence of clinical covariates on population estimates of CL and F of tacrolimus were tested with nonlinear mixed effects models (NONMEM). CL was significantly reduced by elevations in total bilirubin 2.0-9.9 mg/dl (CL * 0.797), bilirubin > or = 10 mg/dl (CL * 0.581), serum creatinine > or = 2 mg/dl (CL * 0.587), grade III/IV graft-versus-host disease (CL * 0.814) and veno-occlusive disease (CL 0.814). No covariates were predictive of oral F. The interindividual variabilities in CL and F were 33% and 44%, respectively. Residual variability was 27.5% and 16.8% at tacrolimus concentrations of 10 microg/l and 20 microg/l, respectively. These models may be used to predict tacrolimus clearance and doses in adult patients following HCT.