A naturally occurring plasmid from Bacillus subtilis, pIM13, codes for constitutively expressed macrolide-lincosamide-streptogramin B (MLS) resistance, is stably maintained at a high copy number, and exists as a series of covalent multimers. The complete sequence of pIM13 has been reported (M. Monod, C. Denoya, and D. Dubnau, J. Bacteriol. 167:138-147, 1986) and two long open reading frames have been identified, one of which (ermC') is greater than 90% homologous to the ermC MLS resistance determinant of the Staphylococcus aureus plasmid pE194. The second reading frame (repL) shares homology with the only long open reading frame of the cryptic S. aureus plasmid pSN2 and is probably involved in plasmid replication. The map of pIM13 is almost a precise match with that of pE5, a naturally occurring, stable, low-copy-number, inducible MLS resistance plasmid found in S. aureus. pIM13 is unstable in S. aureus but still multimerizes in that host, while pE5 is unstable in B. subtilis and does not form multimers in either host. The complete sequence of pE5 is presented, and comparison between pIM13 and pE5 revealed two stretches of sequence present in pE5 that were missing from pIM13. It is likely that a 107-base-pair segment in the ermC' leader region missing from pIM13 accounts for the constitutive nature of the pIM13 MLS resistance and that the lack of an additional 120-base-pair segment in pIM13 that is present on pE5 gives rise to the high copy number, stability, and multimerization in B. subtilis. The missing 120 base pairs occur at the carboxyl-terminal end of the putative replication protein coding sequence and results in truncation of that protein. It is suggested either that the missing segment contains a site involved in resolution of multimers into monomers or that the smaller replication protein causes defective termination of replication. It is concluded that pIM13 and pE5 are coancestral plasmids and it is probable that pIM13 arose from pE5.