Background: Sputum smear microscopy is fast and inexpensive technique for detecting tuberculosis (TB) in high incidence areas but has low sensitivity. Physical and chemical sputum processing along with centrifugation have been found to show promise in overcoming this limitation. Our objective was to compare the sensitivity of smear microscopy obtained with smears made directly from respiratory specimens to those from concentrated specimens. Methods. By active screening, 915 TB suspects were identified from Dhaka Central Jail and sputum specimens were aseptically collected. Direct smears were prepared by taking a small portion of the purulent part of the sputum with a sterile loop. The specimens were then processed by a standard N-acetyl-L-cysteine-NaOH digestion-decontamination method to prepare concentrated specimens. Both smears were then air dried, heat fixed, and stained by the Ziehl-Neelsen staining technique. The stained slides were examined under oil immersion and were graded following International Union Against Tuberculosis and Lung Diseases guidelines. All the specimens were inoculated into Lowenstein-Jensen (L-J) media and culture results were considered as gold standard to calculate sensitivity. Results: Of 915 specimens, 73 (8%) specimens were positive both on direct and concentrated methods, one sample was positive on direct microscopy but was negative on concentrated method. An extra 14 (1.5%) samples were positive on concentrated method which were negative on direct smear. In L-J media 105 specimens were found positive for TB bacilli and of them, 74 (70.5%) and 87 (82.9%) were positive in direct and concentrated smear, respectively. The sensitivity of direct and concentrated smear microscopy was different when using positive culture as the gold standard (71% vs. 83%). Conclusions: The results showed that concentrated technique increases the sensitivity of microscopy up to 12%. Therefore, the national programs in high TB burden countries may consider incorporating the technique into their guidelines at least in the district and higher level laboratories to improve case finding strategy. © 2013 Uddin et al.; licensee BioMed Central Ltd.