With an expected sensitivity (Se) of 96% and specificity (Sp) of 98%, the immunofluorescence antibody test (IFAT) is frequently used as a reference test to validate new diagnostic methods and estimate the canine leihmaniasis (CanL) true prevalence in the Mediterranean basin. To review the diagnostic accuracy of IFAT to diagnose CanL in this area with reference to its Se and Sp and elucidate the potential causes of their variations, a systematic review was conducted (31 studies for the 26-year period). Three IFAT validation methods stood out: the classical contingency table method, methods based on statistical models and those based on experimental studies. A variation in the IFAT Se and Sp values and cut-off values was observed. For the classical validation method based on a meta-analysis, the Se of IFAT was estimated in this area as 89.86% and 31.25% in symptomatic and asymptomatic dogs, respectively. The Sp of IFAT was estimated in non-endemic and endemic areas as 98.12% and 96.57%, respectively. IFAT can be considered as a good standard test in non-endemic areas for CanL, but its accuracy declines in endemic areas due to the complexity of the disease. Indeed, the accuracy of IFAT is due to the negative results obtained in non-infected dogs from non-endemic areas and to the positive results obtained in sera of symptomatic dogs living in endemic areas. But IFAT results are not unequivocal when it comes to determining CanL infection on asymptomatic dogs living in endemic areas. Statistical methods might be a solution to overcome the lack of gold standard, to better categorize groups of animals investigated, to assess optimal cut-off values and to allow a better estimate of the true prevalence aiming information on preventive/control measures for CanL.