We report surveillance for rinderpest virus in wildlife populations in three major ecosystems of East Africa: Great Rift Valley, Somali and Tsavo from 1994 to 2003. Three hundred and eighty wild animals were sampled for detection of rinderpest virus, antigen or genome and 1133 sampled for antibody in sera from Kenya, Uganda, Ethiopia and Tanzania from 20 species. This was done modifying for wildlife the internationally recommended standards for rinderpest investigation and diagnosis in livestock. The animals were selected according to susceptibility and preference given to gregarious species, and populations were selected according to abundance, availability and association with livestock. Rinderpest virus, antigen and/or genome were detected in Kenya; within Tsavo, Nairobi and Meru National Parks. Serological results from 864 animals (of which 65% were buffalo) from the region were selected as unequivocal; showing the temporal and spatial aspects of past epidemics. Recent infection has been only in or peripheral to the Somali ecosystem (in Kenya). Our evidence supports the hypothesis that wildlife is not important in the long-term maintenance of rinderpest and that wildlife are infected sporadically most likely from a cattle source, although this needs to be proven in the Somali ecosystem. Wildlife will continue to be a key to monitoring the remaining virus circulation in Africa. © 2006 Elsevier B.V. All rights reserved