The suitability of the Microbial Identification System (MIS) marketed by Microbial ID (Newark, DE, USA) for routine diagnosis of clinically important mycobacteria was investigated and assessed. Cellular fatty acids were extracted from 1077 stock and recent clinical isolates. They were analyzed using a gas-liquid chromatograph combined with MIS software. The MIS system finally identifies the isolates by comparing their fatty-acid compositions with a standard library for mycobacteria. As the library search usually results in more than one possible match, suitable identification criteria were determined. The stricter these criteria are, the more the percentage of false-positive identifications can be reduced, but at the cost of more cases that remain undecided and require additional analysis. Under conditions similarity index (SI) SI1 > or = 0.4 and SI1-SI2 > or = 0.1, 63% of all isolates were correctly and 6% incorrectly identified.