Rapid and accurate detection of active tuberculosis (TB) cases is one of the most important goal of tuberculosis control programme. For this purpose, new methods are being developed to isolate, serotype and determine the drug resistance of the agent. Xpert MTB/RIF test (CepheidGeneXpert® System, USA) that has been recently developed, is a real-time polymerase chain reaction-based method which detects Mycobacterium tuberculosis complex and resistance of the strain to rifampicin (RIF) from the clinical sample directly within a couple of hours. However, there are not suffi cient data about the performance of that test for extrapulmonary samples and pulmonary samples other than sputum. The aims of this study were to investigate the sensitivity, specifi city, positive and negative predictive values of Xpert MTB/RIF test in detection of M. tuberculosis and the performance in the determination of rifampicin resistance of the isolates from pulmonary and extrapulmonary clinical samples. A total of 2160 clinical samples, in which 1141 (52.8%) were pulmonary and 1019 (47.2%) were extrapulmonary samples, sent to our laboratory between July 2013 to December 2014, were included in the study. Sixty seven of the evaluated samples (3.1%) were positive with microscopy (acid-fast stain; AFS), 116 samples (5.1%) were positive with culture and 98 samples (4.5%) were positive with Xpert MTB/RIF test. When the culture was considered as the reference method, the sensitivity and specifi city of Xpert MTB/RIF test were determined as 73.3% and 99.3%, respectively for all samples; 77.5% and 99.5%, respectively for pulmonary samples and 63.9% and 99.2%, respectively for extrapulmonary samples. Among AFS positive samples, the sensitivity was 100% and specifi city was 66.7%; whereas among AFS negative samples those values were 40.4% and 99.4%, respectively. Among all the samples involved in the study, RIF resistance was determined only in three samples with Xpert MTB/RIF test and that was also proved phenotypically (100% concordance). According to mycobacterial culture results, positive and negative predictive values of Xpert MTB/RIF test were determined as 86.7% and 98.5%, respectively for all samples. Those were determined as 92.5% and 98.3%, respectively for extrapulmonary samples and were determined as 74.2 and 98.7%, respectively for pulmonary samples. According to the results obtained in our study, sensitivity of Xpert MTB/RIF test for extrapulmonary samples was found to be at moderate level; sensitivity of the test was found to be decreased especially in AFS negative samples with less bacilli load. Nonetheless, specifi city of Xpert MTB/RIF test to the agent in all samples was found to be extremely high. In our study, although RIF-resistant strains were detected in few of the samples, Xpert MTB/RIF test could differentiate all resistant and sensitive strains. Additionally, detection of M. tuberculosis and RIF resistance in our laboratory takes approximately 20.96 days with culture, this period decreases to a couple of hours with Xpert MTB/RIF test. Because of the advantages such as being practical, rapid and requiring minimal safety measures, it was concluded that Xpert MTB/RIF test may contribute to rapid diagnosis of TB also in extrapulmonary samples, with the confi rmation of culture method