The usefulness of MTD (Amplified Mycobacterium Tuberculosis Direct Test) for a rapid diagnosis of tuberculosis was evaluated. A total of 400 clinical samples obtained from July, 1995 to June, 1997 were tested by MTD, direct microscopy and culture. The results of MTD and smear/culture were coincident in 387 out of 400 samples. Eight samples (2%) were MTD false-positive (i.e. they were MTD positive but smear and culture negative), and 5 (1.25%) were MTD false-negative (i.e. MTD negative but smear and/or culture positive). Despite a careful review of the clinical data of those patients whose samples showed discrepant results, the reasons of discrepancy were not clear in 2 (0.5%) of the 8 false positives and 3 (0.75%) of the 5 false negatives. In the other cases, the MTD false positives may be accounted for the presence of previous M. tuberculosis infection, the influence of anti-tuberculous medication and so on, and the MTD false negatives are most likely due to the presence of inhibitors (blood, for example) or to the small number of organisms in the specimens. It can be concluded that adequate samples should be obtained, and that MTD should be repeated in case of discrepant results.