Underprediction of in vivo intrinsic clearance (CL int) of unbound drug from human hepatic in vitro systems using physiological extrapolation methodology is accepted as a common outcome. Poulin et al. (2012. J Pharm Sci 101:838-851) recently proposed an approach involving determination of effective fraction unbound in plasma (fu p) based on albumin-facilitated hepatic uptake of acidic/neutral drugs which improved prediction accuracy and precision for 25 drugs highly bound to plasma proteins. This approach includes correction of unbound drug according to the ionisation fraction either side of the plasma membrane based on pH difference. Here, we assessed the proposed method using a larger database of predictions of CL int for 107 drugs involving hepatocytes (89 drugs) and microsomes (64 drugs). The proposed method was similarly effective in minimising average prediction bias (to within twofold), unlike the conventional fu p correction method. However, precision was similar between methods and there was no evidence in the larger database that prediction bias was associated with fu p. Prediction bias for hepatocytes was clearance dependent by either method, indicating important sources of bias from in vitro methodology. Therefore, to progress beyond empirical correction of bias, there is further need of mechanistic elucidation to improve prediction methodology. © 2012 Wiley Periodicals, Inc.