Earlier phylogenetic studies based on the inferred DNA sequences of the polC, rpoB and gyrA genes suggested that strains of the species Bacillus subtilis formed two clusters, indicating the presence two closely related taxa; one contained the laboratory strain 168 and the other the laboratory strain W23. Significant sexual isolation was found between strain 168 and members of the group containing W23, but no sexual isolation was observed between strain 168 and other members of the 168 group. DNA reassociation between the two groups ranged from 58 to 69% and intragroup DNA relatedness ranged from 82 to 100%. Because group 168 strains were highly related to the B. subtilis type strain, they were considered to be bona fide members of the species. About 99.5% sequence identity was observed between the 16S rRNA genes of the 168 and W23 groups. Ribitol and anhydroribitol were principal cell wall constituents of the W23 but not of the 168 group. These observations revealed two closely related but genetically and phenotypically distinct groups within B. subtilis that correspond to two historically important strains. Subspecies distinction is proposed for the 168 and W23 groups, with the names Bacillus subtilis subsp. subtilis subsp. nov. and Bacillus subtilis subsp. spizizenii subsp. nov., respectively. The type strain of the former is NRRL NRS-744T and the latter NRRL B-23049T.