The range of the protozoan parasite Theileria parva, which causes East Coast fever in cattle, has been expanding to countries where it has not previously been detected, as a result of cross-border domestic cattle movement. Countries where T. parva has not previously been observed until recently include Cameroon and South Sudan. This raises the issue of the conservation of the p104 antigen gene, on which the nested PCR assay that is widely used for T. parva surveillance in the blood of infected cattle is based. We sampled 40 isolates from six countries widely distributed across the geographical range of the parasite, including eastern, central and southern Africa, for p104 sequence polymorphism. These included parasites from both domestic cattle and the Cape buffalo (Syncerus caffer) wildlife reservoir. The most frequent allelic variants were present in cattle transmissible isolates from multiple widely separated geographical regions in Zambia, Uganda, Kenya, Tanzania, Rwanda and South Africa. These frequent p104 variants were also present in the three component stocks of the Muguga cocktail used for the infection and treatment live immunisation procedure to control T. parva in the field. Other isolates exhibited unique alleles. This includes some of the p104 sequences from Cameroon, which is outside the known range of the Rhipicephalus tick vector and whose origin is therefore unclear. The nested primer oligonucleotides used to generate the amplicons were universally conserved in cattle-derived parasites and a majority of buffalo-derived isolates across the geographical range of the parasite. However, some rare South African buffalo–derived isolates exhibited one or two mismatches with the primer sequences. It therefore remains possible that some p104 alleles may be so divergent that they do not amplify with the current diagnostic primers and are not detectable in surveys, hence the need for increasing knowledge of genetic heterogeneity of diagnostic targets. There was no evidence for positive selection among those p104 mutations that resulted in residue changes. Importantly, the data indicate that the p104-based PCR detection assay should be effective across the majority of the range of T. parva, and if the one or two mismatches are shown in future to result in the primers annealing less efficiently, then the assay can be further improved by introduction of degenerate bases to enable amplification of the less frequent South African buffalo–derived variant p104 genes.