An epidemiological study of canine leishmaniasis (CanL) was carried out in nine districts of Sfax, in the southern central part of Tunisia. Sera from 250 dogs were tested by two serological methods: the indirect immunofluorescence antibody test and the counter-immunoelectrophoresis. Seven to eight months later, before the next season of transmission, seropositive dogs from the first test were re-examined and a second sampling was performed. Infection status was assessed by serology and by other methods. PCR, in vitro culture and direct examination were applied on blood and other samples (bone marrow, liver, lymph node, spleen and cutaneous biopsies). The seroprevalence of the infection in dogs was 6%. Infection was then confirmed by at least one other method. The PCR is the method which agreed most with serology, all seropositive dogs were found PCR-positive. The sensitivity of the direct examination and the culture was only 33% and 55% respectively as compared with serology. A similar value of seroprevalence has been observed previously in Sousse, in the northern central part of Tunisia. The present report suggests a significant increase of CanL in the Sfax area and confirms that the disease is continuing to move southwards in Tunisia.