Visceral leishmaniasis is a zoonosis in which the dog appears as the main source of infection in urban areas. Its diagnosis is complex and the cytopathological exam is a fast and cheap alternative to parasite direct visualization and its sensitivity can be increased by immunocytochemistry, though with a higher cost. The accuracy of such methods is dependent on the microscopist's experience and therefore, this study evaluated the reliability of such techniques between two observers, from bone marrow aspirates of 50 dogs from an endemic area for the disease. The parasitological culture in Novy-MacNeal-Nicolle medium was used as the reference standard. Among the main findings, the sensitivities obtained by observers I and II were respectively 62.5% and 37.5%, while specificities were 81.1% and 100%. On immunocytochemistry evaluation, the sensitivity was 0% for both evaluators and the specificity 97.3% and 100%. The agreement between evaluators was weak (κ = 0.167) for the cytopathological test and it could not be evaluated for immunocytochemistry, for which there was no detection by the evaluator II. The agreements among the diagnostic methods and the standard reference for the observer I were reasonable (κ = 0.364) for cytopathological examination and bad (κ = -0.041) for immunocytochemistry. For observer II, such agreement could be assessed only for the cytopathological test, being moderate (κ = 0.497). The results point to the possible expertise difference between evaluators, with the evaluator II demonstrating greater experience when interpreting the citopathological test. Although there was the expected sensitivity increase with immunocytochemistry, the technique used in this study was not effective for the diagnosis of infection, regardless of the evaluator.