OBJECTIVES:Bronchoalveolar lavage (BAL) and bronchial washing (BW) are two major methods used to obtain high-quality respiratory specimens from patients with suspected pulmonary tuberculosis (TB) but a sputum-scarce or smear-negative status. We aimed to compare the value of BAL and BW in the diagnosis of TB in such patients. METHODS:We enrolled patients with suspected pulmonary TB but with a sputum-scarce or smear-negative status who were referred for bronchoscopy between October 2013 and January 2016. Participants were randomized into the BAL and BW groups for evaluation. The primary outcome was the diagnostic yield for TB detection. Secondary outcomes included culture positivity, positivity of nucleic acid amplification tests (NAATs) for Mycobacterium tuberculosis and procedure-related complications. RESULTS:A total of 94 patients were assessed and 91 (43 in the BAL group, 48 in the BW group) were analysed. Twenty-one patients (48.8%) in the BAL group and 30 (62.5%) in the BW group had a final diagnosis of pulmonary TB. The detection rate of M. tuberculosis by culture or NAAT was significantly higher in BAL specimens than in BW specimens (85.7% vs 50.0%, p 0.009). The procedure-related complications were hypoxic events, 2/43 (4.7%) in the BAL group and 5/48 (10.4%) in the BW group; and post-bronchoscopic fever, 3/43 (7.0%) in the BAL group and 4/48 (8.3%) in the BW group. DISCUSSION:As long as it is tolerable, BAL rather than BW, should be used to obtain specimens for the diagnosis of pulmonary TB in sputum-scarce or smear-negative cases.