Background: Several nucleic acid amplification techniques are available for detection of Mycobacterium tuberculosis (MTB) in pulmonary and extrapulmonary samples, but insufficient data are available on the diagnostic utility of these techniques in tubercular meningitis where bacilli load is less. The success of final amplification and detection of nucleic acid depends on successful extraction of DNA from the organism. Aims: We performed this study to compare four methods of extraction of MTB DNA from cerebrospinal fluid (CSF) samples so as to select one method of DNA extraction for amplification of nucleic acid from clinical samples. Materials and Methods: Four methods of extracting MTB DNA from CSF samples for testing by real-time polymerase chain reaction (PCR) were compared: QIAGEN R protocol for DNA purification using QIAamp spin procedure (manual), AMPLICOR R respiratory specimen preparation kit, MagNA Pure R kit extraction, combined manual DNA extraction with automated extraction by MagNA Pure R . Real-time PCR was performed on COBAS TaqMan 48 Analyzer R with known positive and negative controls. Results: The detection limit for the combined manual and MagNA Pure R extraction protocol was found to be 100 copies of MTB DNA per reaction as against 1,000 copies of MTB DNA per reaction by the QIAGEN R , AMPLICOR R , and the MagNA Pure R extraction protocol. Conclusion: The real-time PCR assay employing the combination of manual extraction steps with MagNA Pure R extraction protocol for extraction of MTB DNA proved to be better than other extraction methods in analytical sensitivity, but could not detect less than 10 2 bacilli /ml.