OBJECTIVES: To assess the diagnostic validity of laboratory cerebrospinal fluid (CSF) parameters for discriminating between tuberculous meningitis (TBM) and other causes of meningeal syndrome in high tuberculosis incidence settings. METHODS: From November 2009 to November 2011, we included patients with a clinical suspicion of meningitis attending two hospitals in Lima, Peru. Using a composite reference standard, we classified them as definite TBM, probable TBM, and non-TBM cases. We assessed the validity of four CSF parameters, in isolation and in different combinations, for diagnosing TBM: adenosine deaminase activity (ADA), protein level, glucose level, and lymphocytic pleocytosis. RESULTS: One hundred and fifty-seven patients were included; 59 had a final diagnosis of TBM (18 confirmed and 41 probable). ADA was the best performing parameter. It attained a specificity of 95%, a positive likelihood ratio of 10.7, and an area under the receiver operating characteristics curve of 82.1%, but had a low sensitivity (55%). None of the combinations of CSF parameters achieved a fair performance for 'ruling out' TBM. CONCLUSIONS: Finding CSF ADA greater than 6 U/l in patients with a meningeal syndrome strongly supports a diagnosis of TBM and permits the commencement of anti-tuberculous treatment.