in east Africa has been revealed after having remained undetected throughout the period of the JP15 campaign and eight years of PARC. The exact location of this focus is uncertain but surveillance is concentrating on north eastern Kenya and Somalia. It is unlikely that wildlife in Tsavo National Park has acted as a reservoir to maintain the virus in Kenya. Serosurveillance of wildlife populations in north eastern Kenya has not implicated these species in the maintenance of the virus in a silent form. Virus of type 2 lineage isolated from an infected kudu produced only mild disease in Kenyan cattle (Wamwayi and others 1996) while another virus from eland was found to be generally mild in British cattle, although one infected animal died from classical rinderpest (C. Dunn, personal communication). Since this virus is highly virulent in certain wildlife species, wildlife acts as sentinels for the disease. In contrast, the RGK-1 isolate is highly virulent for domestic cattle (Liess and Plowright 1964, Wohlsein and others 1995) and so the pathogenic phenotype of the virus cannot be inferred from its genetic lineage. The endemic focus of the type 2 virus is most likely to be in the border areas between north eastern Kenya and Somalia. In some of these areas rinderpest control is rudimentary or non-existent and it is possible that the virus has persisted unnoticed for over 30 years. Very few sequence changes are needed to alter the virulence of rinderpest virus; the genome of the Plowright vaccine strain differs by less than 0-55 per cent from the virulent virus from which it was derived (Baron and others 1996). Nothing is known concerning the molecular factors which determine the virulence/attenuation of different rinderpest virus strains and it is possible that changes in pathogenic phenotype can occur on passage through different animal species (Plowright 1963). Reversion to the mild form could be a means whereby the virus evades detection for many years. The ability to genetically manipulate morbilliviruses through rescue of live virus from DNA copies of their genomes (Baron and Barrett 1997) means that it will now be possible to address questions concerning virus attenuation and pathogenicity by directly altering virus genes. Only when the molecular basis of pathogenicity is understood will we be able to interpret the sequence data in a scientifically meaningful way. There is major concern over how such a notoriously severe dis-