A Leishmania donovani-complex specific DNA probe was usedto confirm the widespread dissemination of amastigotes in apparently normal skinof dogs with canine visceral leishmaniasis. When Lutzomyia longipalpis were fed on abnormal skin of five naturally infected dogs 57 of 163 (35 per cent) fliesbecame infected: four of 65 flies (6 per cent) became infected when fed on apparently normal skin. The bite of a single sandfly that had fed seven days previouslyon a naturally infected dog transmitted the infection to a young dog from a non-endemic area. Within 22 days a lesion had developed at the site of the infectivebite (inner ear): 98 days after infection organisms had not disseminated throughout the skin, bone marrow, spleen or liver and the animal was still serologically negative by indirect immunofluorescence and dot-enzyme-linked immunosorbent assay. When fed Lu. longipalpis were captured from a kennel with a sick dog known to be infected, 33 out of 49 (67 per cent) of flies contained promastigotes. In contrast only two infections were detected among more than 200 sandflies captured in houses. These observations confirm the ease of transmissibility of L.chagasi from dog to sandfly to dog in Teresina. It is likely that canine VL is the major source of human VL by the transmission route dog-sandfly-human. the Lmet2 DNA probe was a useful epidemiological tool for detecting L. chagasi in sandflies.