The BACTEC 460 radiometric mycobacterial broth culture system has consistently demonstrated faster and increased recovery of Mycobacterium tuberculosis from respiratory specimens of patients with pulmonary tuberculosis than conventional culture methods. We thus questioned whether three sputa were still necessary to definitively diagnose pulmonary tuberculosis if the BACTEC radiometric culture system were in use. We performed a retrospective analysis of 430 sequential respiratory specimens submitted from 143 patients and from which M. tuberculosis had been recovered by in vitro culture and simultaneously assessed the diagnostic yield of acid-fast smear in this same cohort. M. tuberculosis was recovered from the first specimen for 117 (82%) of the 143 patients, from the second for 14 patients (10%; cumulative rate, 92%), and from the third for 12 patients (8%; cumulative rate, 100%). With the exception of those for bronchial brushings, recovery rates of M. tuberculosis were comparable for all respiratory specimen types (expectorated sputum, induced sputum, tracheal aspirates, bronchoalveolar lavage fluids). Only 46 (32%) of these 143 patients had acid-fast bacilli detected in smears; acid-fast bacilli were detected in the first submitted specimen for 44 patients (96%) and in the second for the remaining 2 patients (4%; cumulative rate, 100%). Culture- or smear-positive rates for sequential specimens obtained from AIDS patients were comparable to those for non-AIDS patients. Overall, the diagnostic culture yield of sequentially submitted specimens was not different from previously published studies in which the BACTEC radiometric culture system had not been used. Despite the documented enhanced ability of the BACTEC 460 radiometric mycobacterial culture system to recover M. tuberculosis more often and faster than conventional methods, three sequential respiratory specimens (regardless of type) were still necessary to definitively diagnose pulmonary tuberculosis.