Control of bovine viral diarrhoea (BVD) in Belgium is currently implemented on a voluntary basis at herd level and mainly relies on detection and culling of persistently infected (PI) animals. The present field study was conducted during the winter of 2010/2011 to assess the performances of diagnostic assays used in the testing scheme for BVD as proposed by the two Belgian regional laboratories. Individual blood samples were collected from 4972 animals, and individual samples from the same herd were pooled (maximum of 30 individual samples per pool) and screened for the presence of Bovine Viral Diarrhoea Virus (BVDV)-specific RNA using a commercial real-time RT-PCR test (ADIAGENE). Individual samples from positive pools were then tested in parallel with the same RT-PCR test and with an antigen-capture ELISA test (IDEXX) to detect viremic animals. This study demonstrated that individual results differed according to the type of assay used (P < 0.001): 140 animals (2.8%) were positive by RT-PCR and 72 (1.4%) by antigen-ELISA. A second blood sample was taken 40 days later from 74 PCR positive animals to detect persistent viremia: 17 (23%) of these were still PCR positive and considered to be PI and the 57 that no longer tested positive were assumed to be transiently infected (TI) animals. All PI animals were positive also by antigen-ELISA at both time points. Among TI animals, 10 (16%) were positive by antigen-ELISA at the first but none at the second sampling. A highly significant difference in cycle threshold (Ct ) values obtained by RT-PCR was observed between PI and TI animals. ROC analysis was performed to establish thresholds to confirm with high probability that an animal is PI, based on the result of RT-PCR test performed on a single individual blood sample.