Physiologically based pharmacokinetic (PBPK) modeling of the central nervous system (CNS) provides the opportunity to predict the relevant drug concentrations at the therapeutic target site during preclinical and clinical development. In order to successfully interpret model results, and to provide confidence in the subsequent human predictions, it is essential that an appropriate model structure is chosen at the preclinical stage which takes into account both physiological and drug-specific knowledge. However, the models published to date in the literature show significant variation in the approaches applied by different authors, which can lead to difficulties in the interpretation of model parameter estimates. We aimed to develop a coherent PBPK modeling approach in the rat, which would also be adaptable depending on the quantity and quality of in vivo data obtained during drug development. Based on a sensitivity analysis of the model parameters, and using three CNS drugs as case studies (atomoxetine, acetaminophen, and S 18986), we proposed a decision tree to aid in the appropriate parametrization and structure of the model according to the data available. We compared our parameter estimates to those originally published, and considered the impact of the respective approaches on the mechanistic interpretation of the parameter values. Since the measurement of brain extracellular fluid (ECF) concentrations using microdialysis is not routinely performed in the industrial environment, we also evaluated the bottom-up scaling of in vitro permeability data from the Caco-2 cell line to predict BBB passive permeability in the absence of measured ECF concentrations. Our strategy demonstrates the value of PBPK as a prediction tool throughout the development process of CNS-targeting drugs.