The complete nucleotide sequence of the diphtheria tox228 gene encoding the nontoxic serologically related protein CRM228 has been determined. A comparison of the predicted amino acid sequence with the available amino acid sequences from the wild-type toxin made it possible to deduce essentially the entire nucleotide sequence of the wild-type tox gene. The signal peptide of pro-diphtheria toxin and the putative tox promoter have been identified, a highly symmetrical nucleotide sequence downstream of the toxin gene has been detected; this region may be the corynebacteriophage beta attachment site (attP). The cloned toxin gene was expressed at a low level in Escherichia coli.