Setting: M. tuberculosis isolates were collected from patients attending health clinics in a high incidence urban community and in a low incidence rural setting in South Africa.

Objective: To reconstruct the evolutionary history of a group of closely related M. tuberculosis isolates using IS 6110, DRr and MTB484(1) restriction fragment length polymorphism (RFLP) data.

Design: Mycobacterium tuberculosis isolates containing an average of ten IS 6110 elements, with a similarity index of ⩾65% were genotypically classified by DNA fingerprinting using the IS 6110 derived probes IS-3′ and IS-5′, as well as the DRr and MTB484(1) probes, in combination with Pvu II or Hin fl endonuclease digestion. These RFLP data were subjected to phylogenetic analysis using both genetic distance and parsimony algorithms.

Results: Phylogenetic analysis predicted the existence of two independently evolving lineages, possibly evolving from a common ancestral strain. The topology of the phylogenetic tree was supported by comprehensive bootstrapping and the specific partitioning of DNA methylation phenotypes. The observed difference in the branch lengths of the two lineages may suggest differential evolutionary rates. Isolates collected from different geographical regions demonstrate independent evolution, suggesting that it is highly unlikely that strains have been recently transmitted between the two regions. The number of evolutionary events identified in this strain family differs significantly from that of previously characterized strain families, implying that evolutionary rate may be strain family dependent.

Conclusion: Based on this analysis we propose that the algorithm used to calculate recent epidemiological events should be revised to incorporate the evolutionary characteristics of individual strain families, thereby enhancing the accuracy of molecular epidemiological calculations.