Historically, the focus has been to use in vitro BBB models to optimize rate of drug delivery to the CNS, whereas total in vivo brain/plasma ratios have been used for optimizing extent. However, these two parameters do not necessarily show good correlations with receptor occupancy data or other pharmacological readouts. In line with the free drug hypothesis, the use of unbound brain concentrations (Cu,br) has been shown to provide the best correlations with pharmacological data. However, typically the determination of this parameter requires microdialysis, a technique not ideally suited for screening in early drug development. Alternative, and less resource-demanding methodologies to determine Cu,br employ either equilibrium dialysis of brain homogenates or incubations of brain slices in buffer to determine fraction unbound brain (fu,br), which is subsequently multiplied by the total brain concentration to yield Cu,br. To determine Cu,br/Cu,pl ratios this way, still requires both in vitro and in vivo experiments that are quite time consuming. The main objective of this study was to explore the possibility to directly generate Cu,br/Cu,pl ratios in a single in vitro model of the BBB, using a co-culture of brain capillary endothelial and glial cells in an attempt to mimick the in vivo situation, thereby greatly simplifying existing experimental procedures. Comparison to microdialysis brain concentration profiles demonstrates the possibility to estimate brain exposure over time in the BBB model. A stronger correlation was found between in vitro Cu,br/Cu,pl ratios and in vivo Cu,br/Cu,pl obtained using fu,br from brain slice than with fu,br from brain homogenate for a set of 30 drugs. Overall, Cu,br/Cu,pl ratios were successfully predicted in vitro for 88% of the 92 studied compounds. This result supports the possibility to use this methodology for identifying compounds with a desirable in vivo response in the CNS early on in the drug discovery process.