In 1996, an epidemic outbreak of visceral leishmaniasis (VL) started in Barbar el Fugara, a village in Gedarif State (eastern Sudan). From 1997 to 2000, regular epidemiological studies were carried out in the human population, as well as in mammals and sand flies. In symptomatic patients, 46/69 lymph node, 6/20 post kala-azar dermal leishmaniasis (PKDL) and 1/4 cutaneous cultures in NNN medium were positive. In 69 dogs, 23/79 lymph node cultures were positive. In other mammals (47 rodents, five donkeys, one mongoose and one monkey) spleen and/or blood cultures were negative. Characterization of isolated strains (by starch gel electrophoresis and isoelectrofocusing) identified three zymodemes of Leishmania donovani, two of L. infantum and two of L. archibaldi complexes from patient samples and three zymodemes of L. donovani, three of L. infantum and two of L. archibaldi complexes from dog samples. Five of them were present in both man and dog. For the first time, a strain from a PKDL case was identified as L. infantum, and a child had the same L. infantum zymodeme in VL and in subsequent PKDL. Blood samples from dogs were studied by immunofluorescent antibody test (IFAT). The seroprevalence in dogs was 72.5%, 74.3% and 42.9% in 1998, 1999 and 2000, respectively. By using CDC miniature light traps 12 745 sand flies were collected and then identified. Phlebotomus papatasi (7%) and P. orientalis (5%) were sympatric, mainly inside homes (85% and 75%, respectively). These results, the relative stability of seroprevalence in dogs and the intradomiciliar presence of P. orientalis, known as a vector of VL in Sudan, suggest several hypotheses: (i) man is responsible for the disease in dogs, (ii) the dog is the reservoir of VL, (iii) the dog is an intermediate host between a possible sylvatic cycle and the anthroponotic cycle. More extensive studies are needed to assess the transmission cycle of VL in this area of Sudan.