Human liver microsomes have typically resulted in marked under-prediction of in vivo human intrinsic clearance (CLTint); therefore, the utility of cryopreserved hepatocytes as an alternative in vitro system has become an important issue. In this study, 10 compounds (tolbutamide, diclofenac, S-warfarin, S-mephenytoin, dextromethorphan, bufuralol, quinidine, nifedipine, testosterone, and terfenadine) were selected as substrate probes for CYP2C9, 2C19, 2D6, and 3A4, and the kinetics of metabolite formation (n = 14 pathways) were investigated in three individual lots of cryopreserved hepatocytes and in a pool of human liver microsomes. For the majority of the compounds, lower unbound KM or S50 values were observed in hepatocytes compared with microsomes, on average by 50% over a 200-fold range (0.5-140 μM). Expressed on an equivalent liver weight basis, a good correlation between microsomal and hepatocyte Vmax values was observed for most pathways greater than 5 orders of magnitude (0.16-216 nmol/min/g liver). Unbound hepatocyte CLint (CLint,u) values, when scaled to the whole liver (range 0.38-4000 ml/min/kg), were on average 2.5-fold higher than microsomal CLint,u values, with the exception of tolbutamide and diclofenac, for which lower hepatocellular CLint,u values were observed. Hepatocyte predicted CLint values were compared with human in vivo CLint values, and to supplement our data, in vitro data from cryopreserved hepatocytes were collated from four other published sources. These data show that for 37 drugs, there is, on average, a 4.5-fold under-prediction of the in vivo CLint using cryopreserved hepatocytes, representing a significant reduction in prediction bias compared with human microsomes. Copyright © 2007 by The American Society for Pharmacology and Experimental Therapeutics.