A new commercial assay for the diagnosis of tuberculosis, the BDProbeTec ET Direct Detection assay (Becton Dickinson, USA), was evaluated using 351 respiratory and 372 nonrespiratory specimens. The results were compared to detection of Mycobacterium tuberculosis complex (MTC) by conventional culture. Among the 351 respiratory specimens, MTC bacteria were identified in 150, of which 85 were positive by both microscopy and the assay. Sixty-five specimens culture positive for MTC were microscopy negative; of these, 39 were positive in the assay. All 26 specimens culture positive for nontuberculous mycobacteria (NTM) were negative by the assay. Of 175 specimens culture negative for MTC, 3 were falsely positive by the assay and 1 yielded inhibition. The overall sensitivity and specificity values were 82.7% and 98.5%, respectively. The sensitivity for microscopy-positive and -negative respiratory specimens was 100% and 60%, respectively. After correction for discrepancies, the specificity was 99% compared with notification data. The BDProbeTec ET assay detected 66 of 67 microscopy-positive and 50 of 125 microscopy-negative nonrespiratory specimens. The result for one specimen was inconclusive. All nine specimens containing NTM were negative by the assay. Of 171 specimens culture negative for MTC, 6 were falsely positive by the assay. The overall sensitivity and specificity values obtained with nonrespiratory specimens were 60.7% and 96.7%, respectively. After examining discrepancies by reviewing the patients' histories, the specificity was 98.9%. The sensitivity was 98.5% in microscopy-positive specimens and 40.3% in microscopy-negative specimens. The overall inhibition rate was 0.3%. The BDProbeTec ET assay is a fast, effective, and user-friendly system that can be used for rapid detection of MTC bacteria in respiratory and microscopy-positive nonrespiratory specimens as an important supplement to smear and culture.