RÉSUMÉ
Background. The new guidelines issued by the Revised National Tuberculosis Control Programme for diagnosis of smear-positive tuberculosis recommend examination of only 2 sputum smears. We did a retrospective analysis of data from a designated microscopy centre to ascertain the diagnostic yield of 2 smears and the additional yield provided by the third smear. Methods. Data were obtained from the designated microscopy centre attached to our medical college. A total of 3257 patients with suspected tuberculosis had undergone sputum examination between September 2004 and March 2009. However, only 1762 of them had 3 sputum specimens examined. Data were entered and analysed using SPSS version 11.5. Results. Among the 1762 suspected patients, positivity in any 2 samples was found to be 17.7% while 19% were found to be positive in a single smear. A statistically insignificant association was found between the grading and positivity of the sputum samples using McNemar test. A positive third sample was found in 309 patients. If the first 2 samples were negative, the possibility of missing a third positive sample was 0.4%. Conclusion. Under field conditions, 2 sputum smears are as effective as 3 smears for diagnosing smear-positive tuberculosis.