In December 2008, bird species in two geographically distant holdings were found positive for H5 viruses following the annual Avian influenza serological screening in Belgium. The virological tests performed identified in one holding a low-pathogenic avian influenza (LPAI) virus subtype H5N2, and a H5 LPAI virus was identified by real-time PCR and direct sequencing at the second holding. The first farm was an outdoor mixed holding housing ornamental birds and poultry (n = 6000) and the second a free-range geese breeding farm (n = 1500). No clinical signs or mortalities were reported. Control measures defined by Council Directive 2005/94/EC were followed, including notification to the European Commission via the Animal Disease Notification System and to the World Organization for Animal Health, and poultry were killed, while ornamental bird species were quarantined. Partial sequencing of the H5N2 virus haemagglutinin and neuraminidase N2 gene sequences revealed a close homology to some recent LPAI isolates identified from wild birds in Germany and Italy and from wild birds in Eurasia and Africa, respectively. It is noteworthy that, these two holdings were already H5 positive based on HI test results carried out during the previous serological screening; however, no virus was detected at that time. To have a better understanding of the potential 'silent' circulation of the H5N2 isolate in the field, experimental infections of chickens and turkeys were performed. The low excretion detected might in part explain viral persistence not associated with spread between gallinaceous birds in the same holding, indicating that the H5N2 LPAI isolate was not fully adapted to these two poultry species. Our results highlighted limitations to only using serological screening for the early detection of LPAI in an 'at-risk farm', suggesting that virological and serological monitoring tests be applied simultaneously as a means of testing animals in 'at-risk farms'.