There have been many reports of groups of related Mycobacterium tuberculosis strains described variously as lineages, families or clades. There is no objective definition of these groupings, making it impossible to define relationships between those groups with biological advantages. Here we describe two groups of related strains obtained from an epidemiological study in Tanzania, which we define as the Kilimanjaro and Meru lineages on the basis of IS6110 restriction fragment length polymorphism (RFLP), polymorphic GC rich sequence (PGRS) RFLP and mycobacterial interspersed repeat unit (MIRU) typing. We investigated the concordance between each of the typing techniques and the dispersal of the typing profiles from a core pattern. The Meru lineage is more dispersed than the Kilimanjaro lineage and we speculate that the Meru lineage is older. We suggest that this approach provides an objective definition that proves robust in this epidemiological study. Such a framework will permit associations between a lineage and clinical or bacterial phenomenon to be tested objectively. This definition will also enable new putative lineages to be objectively tested.