We evaluated the sensitivity and specificity of a new semiautomated direct amplification test (DAT), the LCx-MTB, for the diagnosis of pulmonary tuberculosis (TB) and assessed its positive predictive value by focusing on patients with high clinical and radiologic suspicion of pulmonary TB. Respiratory tract specimens from 32 consecutive patients with high suspicion of active pulmonary TB (case patients) and from 204 control patients were cultured for Mycobacterium tuberculosis and tested by LCx-MTB. Sensitivity and specificity of LCx-MTB when compared with culture was, respectively, 80 and 98%. Pulmonary TB was confirmed in the 32 case patients without knowledge of the LCx results: 18 patients were smear- and culture-positive for M. tuberculosis, and all gave at least one specimen that was LCx-positive. Eight patients were smear-negative culture-positive, and seven gave at least one LCx-positive specimen. LCx-MTB was negative in all the specimens obtained from six patients with smear- and culture-negative TB. A positive LCx-MTB result in a smear negative specimen was 100% predictive that at least one of the case patients' specimens would yield M. tuberculosis. As a consequence, knowledge of the LCx-MTB results at the time of specimen collection could have hastened the start of the antituberculosis therapy in three (21%) smear-negative case patients and could have avoided unnecessary invasive diagnostic procedures in four (29%). We conclude that the sensitivity of LCx-MTB in detecting M. tuberculosis DNA in respiratory tract specimens is similar to other DATs, that LCx-MTB is a reliable test for confirmation of TB in smear-positive patients and that LCx-MTB could be beneficial as a diagnostic step in smear-negative patients with a high suspicion of pulmonary TB.