1. Cryopreserved human hepatocytes were used to predict in vivo hepatic clearance (CL(hepatic)) from estimates of in vitro intrinsic clearance (CL' int). 2. (CL' int) was estimated for phenytoin, valproic acid, carbamazepine, theophylline, quinidine and procainamide after their addition to hepatocytes suspended either in human serum or in serum-free media. (CL' int)was estimated from in vitro concentration versus time data fitted to a monoexponential decay model. (CL' int) was estimated from concentrations measured at four time points and from just two-point measures, namely the initial concentration (C(0)) and the final concentration measurement (C(last)). 3. Predicted CL(hepatic) was within twofold of reported in vivo values of CL(hepatic) for all substrates. Moreover, predictions were not significantly different whether derived from hepatocytes suspended in serum or in serum-free medium. 4. Two-point estimates of (CL' int) were just as accurate in predicting CL(hepatic) as were multipoint estimates of (CL' int). 5. Although the data set was limited, the findings suggest that the measurement of the disappearance of xenobiotics from serum or serum-free media in which primary human hepatocytes have been suspended provides a physiologically relevant estimate of hepatic clearance that can be employed early in the drug development process to eliminate xenobiotics with unacceptable clearances.