This study compares smear, growth in Lowenstein-Jensen medium, and in-house polymerase chain reaction (PCR) techniques for the detection of Mycobacterium tuberculosis. A total of 72 specimens from 72 patients with clinical symptoms of tuberculosis, including 70 sputum and two bronchial aspirate samples, were tested in parallel by smear, culture, and in-house PCR techniques. From these, 48 (66.6%) were negative by the 3 methods, 2 (2.8%) were smear positive and negative by culture and in-house PCR, 11 (15.3%) were both smear and culture negative, and in-house PCR positive, 7 (9.7%) were positive by the 3 methods, 2 (2.8%) were positive by smear and culture, and negative by PCR, 2 (2.8%) were positive by culture and PCR, but smear negative. After the resolution of discrepancies in PCR results, the sensitivity and specificity for in-house PCR technique to M. tuberculosis relative to the culture, were 81.8% and 81.9%, respectively. These results confirm that this method, in-house PCR, may be a sensitive and specific technique for M. tuberculosis detection, occurring in both positive and negative smear and negative cultures.