The initiation factor 5A (IF-5A) of archaea and eukaryotes undergoes an unusual post-translational modification consisting of the transformation of a specific conserved lysine residue into the amino acid hypusine. This occurs in a two-step reaction catalysed by the enzymes deoxyhypusine synthase (DHS) and deoxyhypusine hydroxylase. Bacteria do not have IF-5A but only a very distant homologue, the elongation factor P (EF-P). Consequently, all bacteria appeared to also lack genes with significant homology to DHS genes. However, we have carried out BLAST searches and found DHS-like genes in a number of bacterial species. The phylogenetic analysis of these sequences strongly suggests that they have been acquired from archaea by horizontal gene transfer (HGT). Our analysis also suggests, although with weaker support, that a single HGT event from archaea, followed by several HGT between bacterial species, accounts for the patchy distribution of DHS-like genes in bacteria. The activity of these genes in bacteria is enigmatic, since we have not found any evidence of interaction between this protein and the bacterial EF-P. Nevertheless, we cannot discard that it exists, since it appears that the interaction between the DHS and its natural substrate, the IF-5A, is rather weak. This is exemplified by the fact that, in archaea, the complex evolutionary history of the DHS is not paralleled by that of the IF-5A, indicating that these proteins do not follow a perfect co-evolution.