The performances of ELISA assays with different antigen preparations, such as Leishmania amazonensis or L. chagasi lysates and the recombinant antigens rK-39 and rK-26, were compared using sera or eluates from dried blood collected on filter paper to detect anti-Leishmania antibodies in dogs from a visceral leishmaniasis-endemic area in 106 (92.2%) were positive in parasitological exams (skin and/or spleen). These animals were compared to healthy animals (n = 25), negative for IFAT at a titre of 1:40 and parasitological exams. The sensitivities of crude and recombinant antigens were similar and remarkably high for both sera and eluates (97-100%). Specificity was higher than 96% for sera and eluates for different antigens, except for L. chagasi antigen using eluates (88%). Concordance values among the tests were higher either for sera or eluates (J = 0.95-1.00). High concordances were observed between sera and eluates tested with different antigens (kappa = 0.93-0.97) . Crude and recombinant antigens identified different clinical phases of canine leishmaniasis. These results show that eluates could be used in canine surveys to identify L. chagasi infection. Recombinant antigens added little when compared to crude antigen in identifying positive dogs. Cross-reactivity with other diseases whose distribution often overlaps VL-endemic areas is a limitation of crude antigen use however. Key words: canine visceral leishmaniasis -Leishmania (Leishmania) chagasi -diagnosis -recombinant antigensVisceral leishmaniasis (VL) is a zoonosis caused by Leishmania (Leishmania) chagasi and transmitted by the phlebotomine sand fly Lutzomyia (Lutzomyia) longipalpis in Brazil. VL occurs in both rural and urban areas and has become a serious public health problem in several large Brazilian cities in recent decades (Arias et al. 1996). Domestic dogs play an important role in VL maintenance in man-made environments by serving as reservoirs of the parasite (Deane & Deane 1962). In Brazil, canine visceral leishmanisis (CVL) prevalence ranges from 1.9 to 35% in endemic areas (Evans et al. 1990, Nunes et al. 1991, França-Silva et al. 2002. The parasite can be isolated by means of tissue biopsies in 40-50% of dogs with positive immunofluorescence titres (Lanotte et al. 1979, Pozio et al. 1981 as well as asymptomatic animals . Consequently, asymptomatic and symptomatic infected dogs are both infective to the sand fly vectors (Molina et al. 1994).Recommended VL control strategies in Brazil involve systematic treatment of human cases, elimination of seropositive dogs and residual spraying of houses and animal shelters with insecticides (Vieira & Coelho 1998 high proportion of asymptomatic dogs, serological methods are essential for CVL diagnosis. Although the immunofluorescent antibody test (IFAT) is the most widespread diagnostic method, cross-reactivity with other diseases and low sensitivity in detecting asymptomatic dogs are the main limitations of this technique. In addition, IFAT is not readily adaptable to large-scale seroepidemiologi...