The primary structures of the A and B subunits of Shiga toxin and of Shiga-like toxin I (VT1), isolated from the culture supernatants of Shigella dysenteriae 1 and Escherichia coli O157:H7, respectively, were analyzed by Edman degradation of intact proteins and peptides in their digests with trypsin or Achromobacter protease I and also by fast atom bombardment mass spectrometry of the digests. The results indicated that the A and B subunits of Shiga toxin and Shiga-like toxin I have the same primary structures. The identity of their primary structures was confirmed by determining the nucleotide sequence of the gene encoding Shiga-like toxin I cloned from a Shiga-like toxin I converting phage. This nucleotide sequence was different from that reported by Jackson et al. (Microbial Pathogenesis 1987; 2: 147-153), by Calderwood et al. (Proc Natl Acad Sci USA 1987; 84: 4364-8) and by Grandis et al. (J Bacteriol 1987; 169: 4313-9) in one base at position 231, which was found to be adenine instead of thymine, which they reported. The amino acid residue at position 45 from the N-terminus of the A subunit of Shiga-like toxin I deduced from the nucleotide sequence determined in this study is threonine, which corresponds with that found by amino acid sequencing, whereas from previous reports by other investigators it is serine. Edman degradation of the intact A subunit of Shiga toxin indicated that the A subunit was nicked between Ala253 and Ser254 to form A1 and A2 fragments linked by a disulfide bond.