Although, vestibular syndrome is a common neurological presentation, little is known about the diagnostic value of cerebrospinal fluid (CSF) analysis in vestibular syndrome in dogs. Medical records were retrospectively reviewed, and dogs with vestibular disease that had undergone magnetic resonance imaging of the head, CSF analysis and were diagnosed with central or peripheral vestibular syndrome were included. Disorders affecting the central vestibular system included meningoencephalitis of unknown origin (MUO), brain neoplasia, ischaemic infarct, intracranial empyema or metronidazole toxicity. Disorders affecting the peripheral vestibular system included idiopathic vestibular disease, otitis media/interna or neoplasia affecting the inner ear structures. Total nucleated cell concentration (TNCC), total protein concentration (TP) and cytologic assessment were recorded. A total of 102 dogs met the inclusion criteria. The sensitivity and specificity of increased CSF TNCC to differentiate central from peripheral vestibular syndrome was 49% and 90%, while the sensitivity and specificity of increased TP was 58% and 39%, respectively. The TNCC and TP in dogs with MUO were significantly higher than in dogs with idiopathic vestibular disease (p=0.000 and p=0.004). MUO was associated with lymphocytic pleocytosis, while idiopathic vestibular disease and ischaemic infarct were associated with the presence of activated macrophages or normal cytology (p=0.000). Although consistent CSF abnormalities were observed in dogs with MUO, CSF analysis did not allow reliable differentiation between central and peripheral vestibular syndrome. CSF analysis is not reliable as the sole diagnostic technique in dogs with vestibular disease.