Children and adults with malignant diseases have a high risk of prevalence of the tumor in the central nervous system (CNS). As prophylaxis treatment methotrexate is often given. In order to monitor methotrexate exposure in the CNS, cerebrospinal fluid (CSF) concentrations are often measured. However, the question is in how far we can rely on CSF concentrations of methotrexate as appropriate surrogate for brain target site concentrations, especially under disease conditions. In this study, we have investigated the spatial distribution of unbound methotrexate in healthy rat brain by parallel microdialysis, with or without inhibition of Mrp/Oat/Oatp-mediated active transport processes by a co-administration of probenecid. Specifically, we have focused on the relationship between brain extracellular fluid (brainECF) and CSF concentrations. The data were used to develop a systems-based pharmacokinetic (SBPK) brain distribution model for methotrexate. This model was subsequently applied on literature data on methotrexate brain distribution in other healthy and diseased rats (brainECF), healthy dogs (CSF) and diseased children (CSF) and adults (brainECF and CSF). Important differences between brainECF and CSF kinetics were found, but we have found that inhibition of Mrp/Oat/Oatp-mediated active transport processes does not significantly influence the relationship between brainECF and CSF fluid methotrexate concentrations. It is concluded that in parallel obtained data on unbound brainECF, CSF and plasma concentrations, under dynamic conditions, combined with advanced mathematical modeling is a most valid approach to develop SBPK models that allow for revealing the mechanisms underlying the relationship between brainECF and CSF concentrations in health and disease.