Nucleotide sequences specific for a range of Mycobacterium species were defined by computer-assisted sequence comparisons of small subunit ribosomal RNA. A polymerase chain reaction-based sequencing strategy was used to demonstrate that the 16S rRNA sequence can be used for the rapid identification of mycobacterial isolates. Identification at the species level can be obtained within 2 d, requiring less than 10,000 bacteria. This procedure reliably differentiates Mycobacterium spp. which are difficult to identify by classical methods, such as M. malmoense, M. szulgai and M. flavescens.