Markov chain analysis of the Haemophilus influenzae Rd genome reveals striking under-representation of three palindromic tetranucleotide strings (CCGG, GGCC and CATG), accompanied by over-representation of six tetranucleotide strings that are derived from the former by exchanging strand location of the two residues making up a G/C nucleotide pair at the terminal palindrome position. Constraints are outlined for a molecular model able to explain the phenomenon as the result of sequence-targeted, enzyme-driven G/C to C/G transversion mutagenesis. Possible participation in the process by components of known DNA mismatch repair or restriction/modification systems (in particular, cytosine methylation) is discussed. The effect widens the spectrum of enzyme-driven, specific mutagenesis beyond the formerly described C/G to T/A transition (VSP repair of Escherichia coli). Potential evolutionary benefits of enzymatic pathways of specific mutagenesis can be envisioned.