BACKGROUND:Smear-negative tuberculosis occurs more frequently in human immunodeficiency virus (HIV)-infected patients than in non-HIV-infected patients. Besides, there are substantial numbers of patients who cannot produce sputum, making the diagnosis of pulmonary tuberculosis (PTB) difficult.AIMS:To evaluate the relative yield of pre- and post-bronchoscopy sputum and bronchoalveolar lavage (BAL) in ‘sputum smear’-negative, HIV-positive patients.SETTINGS:A tertiary care referral hospital in Addis Ababa.MATERIALS AND METHODS:Acid-fast stain (AFS) using the concentration technique was done on 85 pre-bronchoscopy sputum and 120 BAL samples. Direct AFS was done on all BAL and 117 post-bronchoscopy sputum samples. Culture for Mycobacterium tuberculosis (MTB) was done for all sputa and BAL.RESULTS:MTB was isolated from 26 (21.7%), 23 (19.7%) and 13 (15.3%) of BAL, post- and pre-bronchoscopy sputum cultures respectively. AFS on pre-bronchoscopy sputum using concentration technique and direct AFS on BAL together detected 11 (41%) of the 27 culture-positive cases. In patients who could produce sputum, the sensitivity of pre-bronchoscopy sputum culture (13/85, 15.3%) was comparable to BAL (12/85, 14%) and post-bronchoscopy sputum (12/85, 14%). In patients who could not produce sputum, however, both BAL (12/35, 40%) and post-bronchoscopy sputum (12/32, 31.4%) detected significantly more patients than those who could produce sputum (P=0.002, P=0.028 respectively).CONCLUSION:In HIV-infected patients, AFS by concentration method on pre-bronchoscopy sputum and direct AFS on BAL in patients who cannot produce sputum are the preferred methods of making a rapid diagnosis. BAL culture seems to add little value in patients who can produce sputum; therefore, bronchoscopy should be deferred under such circumstances.