Background. Tuberculous meningitis (TBM) is a severe complication of tuberculosis predominantly affecting young children. Early treatment is vital to prevent morbidity and mortality, emphasizing the importance of early diagnosis. The lack of sensitive methods for early diagnosis is the most common cause of delay. Attempts have been made to develop simplified tests for tuberculosis, but their diagnostic power remains poor. The clinical picture of TBM is mainly driven by the host's immune response to Mycobacterium tuberculosis; therefore, identification of disease-specific biomarkers may have diagnostic and therapeutic value and improve our understanding of its pathogenesis. Methods. We investigated disease-specific biomarkers of childhood TBM in a cohort of children aged 3 months-13 years with symptoms and signs suggestive of meningitis. Cerebrospinal fluid (CSF) and serum from 56 patients with and 55 patients without TBM were assessed for 28 soluble mediators. Results. Unsupervised hierarchical clustering analysis revealed a disease-specific pattern of biomarkers for TBM relative to other types of meningitis. A biomarker-based diagnostic prediction model for childhood TBM based on CSF concentrations of interleukin 13 (cutoff value, 37.26 pg/mL), vascular endothelial growth factor (cutoff value, 42.92 pg/mL), and cathelicidin LL-37 (cutoff value, 3221.01 pg/mL) is presented with a sensitivity of 0.52 and a specificity of 0.95. Conclusions. These data highlight the potential of biosignatures in the host's CSF for diagnostic applications and for improving our understanding of the pathogenesis of TBM to discover strategies to prevent immunopathological sequelae.