Objective: Rapid and accurate diagnosis of Mycobacterium tuberculosis which is one of the first ten mortal microorganisms all over the world is very important to start treatment rapidly and to prevent new infections. Our aim is to evaluate the diagnostic performance of Xpert MTB/RIF assay for clinical samples in Kütahya where TB incidence remains under mean values for Turkey. Methods: Clinical samples taken from patients suspected of pulmonary and extrapulmonary tuberculosis were tested with Xpert MTB/RIF assay without any prior processing; and the same samples after NALC-NaOH homogenisation, decontamination or without; depending on the sample type were tested with EZN staining, inoculation on LJ medium,. Performances of culture and Xpert MTB/RIF test were determined in respiratory/ nonrespiratory and smear negative/positive samples; categories of AFB and Xpert MTB/RIF were compared and agreement between diagnostic methods was calculated. Results: Out of 209 patients 26 (12,4%) were accepted as tuberculosis (76.9%, male; 50%, <45 years) after clinical, radiological and microbiological evaluation. Positivity of respiratory and nonrespiratory samples with culture and/or molecular method were 13.2% and 12.5%, respectively. In smear negative respiratory samples, sensitivity and specifity of Xpert MTB/RIF were 84.6% and 100%, in nonrespiratory samples 60% and 100%; in all smear positive samples, sensitivity and specifity were 100%. Smear microscopy had lowest sensitivity for nonrespiratory samples and highest sensitivity for bronchial lavage; sensitivity of Xpert MTB/RIF was similar to culture for respiratory samples and was more sensitive for nonrespiratory samples; agreement in categories of AFB/Xpert MTB/RIF was highest in respiratory samples. Conclusion: In conclusion establishment of Xpert MTB/RIF system together with conventional diagnostic methods will contribute to rapid and reliable results for laboratories with low sample burden in failures of economic and physical conditions.