The advantages of Mycobacteria Growth Indicator Tube (MGIT) system were analyzed and compared to Löwenstein-Jensen (LJ) and Stonebrink (S) solid media when searching for a fast method to diagnose tuberculosis and mycobacterioses, which should be easy to perform in laboratories and of non-invasive reading. All nonsterile specimens were pretreated with Petroff method. A total of 191 specimens were processed (among which 152 were pulmonary and 39 extrapulmonary). Twenty-nine tested positive by one of the methods employed. The rate of recovery of smear positive specimens (ED+) was of 92% with LJ/S, and 85% with MGIT. The mean time to detect a positive result was 18.7 days with MGIT, and 20.6 days with LJ/S. The rate of recovery of smear negative specimens (ED-) was 88.2% with LJ, 70.6% with S, and 23.5% with MGIT. The mean time to detect a positive result were 38.5; 26.5 and 29 days, respectively. During this experiment, MGIT did not seem to have any advantage over the traditional methods, particularly when working with samples containing a low number of bacilli (p < 0.05). It is obviously necessary to make a comparative study with a large amount of cases which might support this observation and to analyze the influence of the reagents employed in the pre-treatment of the specimens on the MGIT system's efficacy to detect mycobacteria.