This study was based on the need to employ a sensitive and specific method with samples that could be easily collected for diagnosing dogs infected with Leishmania infantum. To this end, we used real time-PCR (qPCR) to assess the value of the oral swab (OS) in detecting infected sick dogs (SD; n=62), including, for the first time, the analysis of apparently healthy infected dogs (AD; n=30), both from endemic areas for visceral leishmaniasis (VL). For comparison, we also evaluated the performance of the conjunctival swab (CS), blood (BL), lymph node (LN) and serology. We detected the presence of Leishmania DNA in the oral cavity in 62 out of the 92 dogs studied. The OS positivity (67.4%) was equivalent to the CS (68.5%) (p>0.05), higher than BL (52.2%) (p≤0.05), and lower than LN (84.8%) (p≤0.05). OS and CS performed well in SD dogs (82.3% and 83.9%, respectively) but not in AD dogs (36.7% for both samples). BL showed the lowest positivity (52.2%) and provided equivalent results between AD (60.0%) and SD (48.4%) dogs (p>0.05). LN yielded the highest positivity (84.8%), and it was also higher in the SD population (93.5%) compared to the AD population (66.7%) (p≤0.05). Parasite load was high in LN, moderate in OS and CS, and low in BL, showing the relationship between the levels of parasitism and the positivity rates found in these samples. Serology was positive in 82.2% of the SD group and in 70% of the AD dogs (p>0.05). Among the 20 seronegative dogs, seven (35%) were positive in either OS or CS, and 12 (60%) were positive when both noninvasive samples were jointly considered. The OS/CS combination resulted in a significant increase of positivity (p≤0.05) for the AD dogs (from 36.7% to 63.4%), as well as OS/serology (80%) and OS/CS/serology (83.4%). For the SD population, positivity reached up to 95.2% with the same combinations, showing that combination of samples and/or tests is required for the identification of dogs infected with L. infantum and that the OS and CS combination based on qPCR notably improves the detection of both AD and SD dogs. In conclusion, OS proved to be a suitable sample for the molecular diagnosis of infected dogs with clinical signs of VL, but not for dogs with inapparent infection. For these, we recommend the combination of OS results with CS and/or serology in order to reach relevant positivity for L. infantum. Finally, another advantage of using OS or both noninvasive samples is the increased likelihood of diagnosing seronegative dogs.