The availability of the dnaJ1 gene for identifying Mycobacterium species was examined by analyzing the complete dnaJ1 sequences (approximately 1200 bp) of 56 species (54 of them were type strains) and comparing sequence homologies with those of the 16S rRNA gene and other housekeeping genes (rpoB, hsp65). Among the 56 Mycobacterium species, the mean sequence similarity of the dnaJ1 gene (80.4%) was significantly less than that of the 16S rRNA, rpoB and hsp65 genes (96.6%, 91.3% and 91.1%, respectively), indicating a high discriminatory power of the dnaJ1 gene. Seventy-one clinical isolates were correctly clustered to the corresponding type strains, showing isolates belonging to the same species. In order to propose a method for strain identification, we identified an area with a high degree of polymorphism, bordered by conserved sequences, that can be used as universal primers for PCR amplification and sequencing. The sequence of this fragment (approximately 350 bp) allows accurate species identification and may be used as a new tool for the identification of Mycobacterium species.