Isolated perfused trout livers were used to evaluate in vitro-in vivo metabolism extrapolation procedures for fish. In vitro depletion rates for 6 polycyclic aromatic hydrocarbons (PAHs) were measured using liver S9 fractions and extrapolated to the intact tissue. Predicted hepatic clearance (CLH) values were then compared with values exhibited by intact livers. Binding in liver perfusates was manipulated using bovine serum albumin (BSA) and was characterized by solid-phase microextraction. Additional studies were conducted to develop binding terms (f U; calculated as the ratio of unbound fractions in liver perfusate [f U,PERF] and the S9 system [f U,S9]) used as inputs to a well-stirred liver model. Hepatic clearance values for pyrene and benzo[a]pyrene, predicted by extrapolating in vitro data to the intact tissue, were in good agreement with measured values (< 2-fold difference). This can be partly attributed to the rapid rate at which both compounds were metabolized by S9 fractions, resulting in perfusion-limited clearance. Predicted levels of CLH for the other PAHs underestimated observed values although these differences were generally small (< 3-fold, except for naphthalene). Setting f U = 1.0 improved clearance predictions at the highest tested BSA concentration (10mg/ml), suggesting that trout S9 fractions exhibit lower levels of intrinsic activity than the intact tissue or that the full binding assumption (ie, f U = f U,PERF/f U,S9) underestimates the availability of hydrophobic substrates to hepatic metabolizing enzymes. These findings provide qualified support for procedures currently being used to predict metabolism impacts on chemical accumulation by fish based on measured rates of in vitro activity.