A method developed in the USA (2, 6, 8) for the culturing, identification and differentiation of mycobacteria by means of a radiometric method (Bactec) was compared with a conventional method by examining 802 specimens received. Obviously better results were obtained with the new method: The number of positive cultures was 102 (12.7%) for Bactec against 87 (10.8%) for the conventional method. When using the radiometric method, the contamination rate (5.1%) was higher than for the conventional method (3.1%). It would, however, seem that this disadvantage can be offset by an increase in the alkali concentration during pretreatment of the specimens. After elimination of all paired samples one or both specimens were found to be contaminated, 743 specimens remained for direct comparison. Of these, 101 (13.6%) were positive when the Bactec method, and 84 (11.3%), when the conventional method was used. The superiority of the new method was most obvious with sputum specimens: 14.5% were positive when Bactec, and 12.2%, when the conventional method was used. For the Bactec method, the mean period until positive results could be recognized by daily readings was 15 days against 28 days for the conventional method with weekly readings. Sensitivity testing can be completed within 8 days. Owing to the costs of the radiometric method, it is recommended to limit its use of defined situations.