Highly lipophilic compounds are often encountered in the early stages of drug discovery. The apparent permeability (Papp) of these compounds in Caco-2 cell could be underestimated because of considerable retention by the Caco-2 monolayer and non-specific binding to transwell surface. We have utilized a general approach for the determination of permeability of these compounds, which includes the addition of 1-5% DMSO in the apical (AP) and 4% bovine serum albumin (BSA) in the basolateral (BA) side. Two highly lipophilic and highly protein bound Schering compounds, SCH-A and SCH-B, exhibited poor recovery and low Papp in the conventional Caco-2 system that included 1% DMSO in the AP and BA sides. In contrast, both compounds were well absorbed in cynomolgus monkeys. Inclusion of BSA (up to 4%) in the BA side provided necessary absorptive driving force similar to in vivo sink conditions improving both recovery and Papp of these compounds as well as progesterone, a model highly lipophilic and highly protein bound compound. Whereas, the recovery and Papp of mannitol (high recovery, low permeability) and propranolol (high recovery, high permeability) remained unaffected. The presence of 4% BSA increased Papp of SCH-A, SCH-B, and progesterone by five-, four-, and three-fold, respectively. We also compared this approach with a second, based on the disappearance of the compound from the AP side, which resulted in a reasonable estimate of the permeability (23.3x10(-6) cm/s) for SCH-A. The results demonstrated that the reliable estimates of permeability of highly lipophilic compounds that are subjected to considerable retention by the cell monolayer and exhibit non-specific binding are obtained by the addition of BSA to the BA side.