The objective of the analysis described herein is to examine the in vitro/in vivo relationship of estimated bioavailability values and also the applicability of the estimated in vitro bioavailability to lead candidate selection in drug discovery. To this end, in vitro ADME data from screening assays as well as in vivo rat pharmacokinetic (PK) data were compiled for 140 compounds across therapeutic areas from the Pfizer library in Ann Arbor. The compounds span a broad range of structural types, including neutral, basic, and acidic compounds. Solubility and Caco-2 permeability data from in vitro ADME screening were used to calculate the fraction dose absorbed (FDp) using the physiologically based IDEA model. In vitro metabolic stability (t(1/2)) from human and rat liver microsomal incubations was converted to an in vitro intrinsic clearance value (CL(int)'), which was then scaled up to reflect in vivo clearance (CL) and hepatic extraction as described by Obach et al. [J. Pharmcol. Exp. Ther. 283 (1997) 46]. Subsequently, the in vitro/in vivo relationship between the measured bioavailability (F(obs)) in rats and the estimated bioavailability (F(est)) from FDp and predicted CL values was examined. The observed data suggest that compounds with low estimated in vitro bioavailability (F(est)<15%) are more likely to have low in vivo bioavailability (F(obs)<30%). Therefore, the present study indicates that in vitro estimation of bioavailability is an efficient tool to eliminate compounds having low bioavailability prior to in vivo characterization and therefore can be used to reduce attrition due to poor ADME properties in drug development.