The purpose of this investigation was to evaluate the Amplified Mycobacterium tuberculosis Direct Test (AMTDT) for the diagnosis of pulmonary tuberculosis (TB). Sixhundred and forty-six sputum samples were analysed by microscopy for acid-fast bacilli, by culture for mycobacterial growth, and by AMTDT for the presence of M. tuberculosis complex rRNA. If there were discrepant results, information as to whether the patient had a history of TB was obtained. The sensitivity was 85.7% (smear positive 100% and smear negative 71.4%, respectively) and the specificity was 96.8% compared with culture. By retesting, 12 of 13 false-positive samples could be divided into two major categories. The samples in one category were retest positive, had highly positive results, and the patients had previous culture-proven TB. The samples in the other category had less positive AMTDT results, were retest negative, and the patients were not notified as having TB. This investigation shows that AMTDT is a sensitive supplementary method for rapid detection of M. tuberculosis complex in respiratory specimens, that there is a risk of contamination which has to be monitored closely, and that positive samples should be retested.