Phage lysin was evaluated as a substitute for antibiotics in sputum samples processed by a modified Petroff's method for the detection of Mycobacterium tuberculosis with the MGIT 960 system. One hundred and fifty sputum samples were processed, inoculated onto two slopes of Lowenstein-Jensen medium, and divided in to two aliquots of 0.5 mL each. One aliquot was added to 7 mL of MGIT medium containing polymyxin B, amphotericin B, nalidixic acid, trimethoprim and azlocillin (PANTA) (MGIT-PANTA) and the other was added to 7 mL of MGIT medium containing 0.8 mL of lysin (MGIT-Lysin). The samples were randomized and incubated at 37°C in the MGIT 960 system. The sensitivity and specificity of MGIT-Lysin were 97% and 88%, respectively, as compared with MGIT-PANTA. The average times to detection with MGIT-Lysin and MGIT-PANTA were 9.3 and 8.6 days, respectively. The rate of contamination with MGIT-PANTA and MGIT-Lysin were 16% and 7.3%, respectively. Phage lysin can be substituted for antibiotics in processed sputum samples for the detection of M. tuberculosis.