Microbiologic confirmation of pulmonary tuberculosis among patients whose sputum smear is negative is increasingly important because of greater incidence among immunocompromised hosts and emergence of drug-resistant strains. We prospectively compared sputum induction to fiber-optic bronchoscopy in the diagnosis of such patients. Consecutive patients referred for investigation of possible active pulmonary tuberculosis underwent sputum induction with hypertonic saline delivered by an ultrasonic nebulizer between 2 and 48 h before transnasal fiber-optic bronchoscopy. All specimens were examined for acid-fast bacilli with fluorescent microscopy and cultured for mycobacteria. Clinical information was abstracted from patient records, and X-rays were reviewed by two blinded readers. Among 101 participants, sputum induction was well-tolerated without complications and provided adequate samples in 93. Sensitivity of direct acid-fast bacilli smear of specimens from both techniques was low. Sensitivity and negative predictive value of culture from bronchoscopy specimens was 73% and 91% compared with 87% and 96%, respectively, for sputum induction when a specimen was obtained. Direct costs for bronchoscopy totaled Canadian $187.60 compared with Canadian $22.22 for sputum induction. Sputum induction was well-tolerated, low-cost, and provided the same, if not better, diagnostic yield compared with bronchoscopy in the diagnosis of smear-negative pulmonary tuberculosis.