Despite several data on influenza infection in dogs, the first natural outbreak of canine influenza virus, closely related to H3N8 equine subtype, dates back to 2004 in Florida. Subsequent studies highlighted the role of dogs in adaptation of H5N1 to mammals and the susceptibility of dogs to different subtypes of influenza. A prevalence study was carried out on 562 sera collected from pet and kennel dogs in the south of Italy. A c-ELISA test was employed and c-ELISA-positive, c-ELISA-doubtful and random c-ELISA-negative samples were also tested in subtype- specific HI test using H3N8 and H3N2 strains. c-ELISA detected a positivity of 3.56%. HI performed with the H3N8 revealed 2 positive samples and when performed with the H3N2, HI revealed 47 positive samples. c-ELISA proved to be a sensitive and specific technique. HI is a specific method only when the test antigen is homologous to the circulating virus and, because non-specific-hemagglutination inhibitors may be present in dog sera, false positives can result. The study emphasizes that due to their close contact with humans dogs must be a target for testing. Furthermore because it remains to be determined how long antibodies to influenza virus persist in canine sera, the observed prevalence might be underestimated.