Background: Pathological cerebral conditions may manifest in altered composition of the cerebrospinal fluid (CSF). Although diagnostic CSF analysis seeks to establish pathological disturbances in the brain proper, CSF is generally sampled from the lumbar compartment for reasons of technical ease and ethical considerations. We here aimed to compare the molecular composition of CSF obtained from the ventricular versus the lumbar CSF compartments to establish a relevance for employing lumbar CSF as a proxy for the CSF bathing the brain tissue. Methods: CSF was collected from 46 patients with idiopathic normal pressure hydrocephalus (iNPH) patients during their diagnostic workup (lumbar samples) and in connection with their subsequent CSF diversion shunt surgery (ventricular samples). The mass-spectrometry-based proteomic profile was determined in these samples and in addition, selected biomarkers were quantified with ELISA (S100B, neurofilament light (NfL), amyloid-β (Aβ₄₀, Aβ₄₂), and total tau (T-tau) and phosphorylated tau (P-tau) forms). The latter analysis was extended to include paired porcine samples obtained from the lumbar compartment and the cerebromedullary cistern closely related to the ventricles. Results: In total 1231 proteins were detected in the human CSF. Of these, 216 distributed equally in the two CSF compartments, whereas 22 were preferentially (or solely) present in the ventricular CSF and four in the lumbar CSF. The selected biomarkers of neurodegeneration and Alzheimer’s disease displayed differential distribution, some with higher (S100B, T-tau, and P-tau) and some with lower (NfL, Aβ₄₀, Aβ₄₂) levels in the ventricular compartment. In the porcine samples, all biomarkers were most abundant in the lumbar CSF. Conclusions: The overall proteomic profile differs between the ventricular and the lumbar CSF compartments, and so does the distribution of clinically employed biomarkers. However, for a range of CSF proteins and biomarkers, one can reliably employ lumbar CSF as a proxy for ventricular CSF if or a lumbar/cranial index for the particular molecule has been established. It is therefore important to verify the compartmental preference of the proteins or biomarkers of interest prior to extrapolating from lumbar CSF to that of the ventricular fluid bordering the brain.