The maintenance and transmission of human visceral leishmaniasis in an endemic area usually requires a mammalian reservoir. Though universal reservoir elimination has previously been effective in controlling this disease in countries where the primary reservoir is the dog, selective elimination would be preferable. To guide this selection process, we performed a prospective, single, blind, cohort study evaluating the dot-enzyme-linked immunosorbent assay (dot-ELISA) in 130 canines from both endemic and nonendemic areas. The results were compared with blinded bone marrow aspirate examination and physical assessment of the animals. Using visualization of amastigotes on bone marrow examination as a priori evidence of infection, the dot-ELISA was found to be highly sensitive (97%) and specific (100%). In contrast, the physical evaluation had remarkably low sensitivity and specificity. The dot-ELISA is an excellent test for detection of the canine reservoir of Leishmania. Because it is simple to perform, inexpensive, and highly accurate, it may help control this debilitating illness by facilitating selective canine elimination.