GenoType MTBDRplus is a molecular assay for detection of Mycobacterium tuberculosis and drug resistance. Assay performance as applied directly to consecutive unselected sputum samples has not been established. The objective of this study was to determine the accuracy of the MTBDRplus test for direct detection of M. tuberculosis (in sputum) and for drug resistance in consecutively submitted sputum samples. In this cross-sectional study in South Africa, one sputum specimen from each person suspected of having pulmonary tuberculosis was tested by smear microscopy, direct MTBDRplus, and Mycobacterial Growth Indicator Tube (MGIT) culture with MGIT drug susceptibility testing. MGIT results were the reference standard. We tested 2,510 sputum samples, and 529 (21.1%) were positive for M. tuberculosis by MGIT. Direct MTBDRplus identified M. tuberculosis in 256 of 529 specimens (sensitivity, 48.4%; 95% confidence interval [CI], 44.1, 52.7). The sensitivity of MTBDRplus for M. tuberculosis detection by sputum smear status was as follows: smear negative, 13.7% (95% CI, 9.8, 18.4); smear scanty, 46.2% (95% CI, 19.2, 74.9); smear 1+, 69.1% (95% CI, 55.2, 80.9); smear 2+, 86.3% (95% CI, 73.7, 94.3); smear 3+, 89.8% (95% CI, 83.7, 94.2). Direct MTBDRplus testing was negative for 1,594/1,612 sputum samples that were culture negative for M. tuberculosis (specificity, 98.9%; 95% CI, 98.2, 99.3). For specimens positive for M. tuberculosis by MTBDRplus, this assay's sensitivity and specificity for rifampin resistance were 85.7% (95% CI, 57.2, 98.2) and 96.6% (95% CI, 93.2, 98.6) and for isoniazid resistance they were 62.1% (95% CI, 42.3, 79.3) and 97.9% (95% CI, 94.8, 99.4). For sputum testing, the sensitivity of MTBDRplus is directly related to the specimen's bacillary burden. Our results support recommendations that the MTBDRplus test not be used for direct testing of smear-negative or paucibacillary sputum samples.