A study was performed to diagnose tuberculosis by smear, culture, and nucleic acid amplification. The study was comprised of two independent arms. Each arm used a different specimen processing method; in one arm, all specimens were processed with N-acetyl-l-cysteine-sodium hydroxide, and in the other arm, all specimens were processed with C(18)-carboxypropylbetaine and lytic enzymes. In each arm, all processed sediments were split for analysis by auramine smear, by culture using the MB/BacT liquid culture system and solid media, and by nucleic acid amplification using the COBAS AMPLICOR MTB test. In the N-acetyl-l-cysteine-sodium hydroxide arm, 1,468 specimens were analyzed: 65 were smear positive; 88 and 42 were culture positive for Mycobacterium tuberculosis and nontuberculous mycobacteria, respectively; and 103 were PCR positive. Relative to cultures positive for M. tuberculosis, the sensitivity and specificity of the smear were 68.2% and 99.6%, respectively, and those of PCR were 75.0% and 97.3%, respectively. In the C(18)-carboxypropylbetaine study arm, 1,423 specimens were analyzed: 44 were smear positive; 82 and 31 were culture positive for M. tuberculosis and nontuberculous mycobacteria, respectively; and 91 were PCR positive. The sensitivity and specificity of the smear were 48.8% and 99.7%, respectively, and those of PCR were 78.0% and 98.0%, respectively. When the two arms were compared, C(18)-carboxypropylbetaine specimen processing significantly increased the number of smear-negative and culture-positive specimens and significantly increased the PCR sensitivity among this same group of specimens while at the same time significantly reducing the inhibition rate.