Diazepam, a drug of high intrinsic clearance, was studied in the in situ rat liver dually perfused with Krebs-bicarbonate buffer containing human serum albumin (HSA; 0-1%) and unlabeled diazepam (1 mg/l) under constant hepatic arterial (3 ml/min) and portal venous (PV; 12 ml/min) flow rates. Events after a unit impulse (using [C-14] diazepam) and at steady state (using unlabeled diazepam) were evaluated. In the absence of HSA the fractional effluent recovery (F) after hepatic arterial infusion (0.046 +/- 0.013) was about twice that after PV infusion (0.019 +/- 0.006). With HSA present, regardless of input route, F increased as unbound diazepam fraction in perfusate decreased (e.g., for PV, F = 0.58 +/- 0.05 and 0.69 +/- 0.02 for unbound diazepam fraction values of 0.18 +/- 0.01 and 0.037 +/- 0.01 at 0.25% and 1% HSA). The effluent [14C] diazepam profile was also dependent upon HSA. On decreasing HSA from 1 to 0.25% the early sharp peak (at 12-20 s) was replaced by a flatter unimodal profile with a later peak (at 60-80 s). Comparison of estimated effective permeability-surface area product to perfusate flow ratios (4.4 for 1% HSA and 21 for 0.25% HSA) indicated a shift from a perfusion rate-limited uptake with 0.25% HSA to one intermediate between permeability and perfusion at 1% HSA. Recognizing that orally absorbed drug enters the liver only via PV and i.v. drug via both vascular routes, this study emphasizes the difference in hepatic extraction of compounds depending on route of input, and the role that alteration in perfusate binding has on hepatic drug disposition.