The aim of this study was to examine whether it is possible to predict the presence of persistently infected (PI) animals with bovine viral diarrhoea virus on dairy farms in The Netherlands, based on a few blood samples of the herd, possibly in combination with a bulk milk test for antibodies. In 25 herds with, and 24 herds without, PI animal(s) the probabilities of obtaining at least x antibody positive animals out of a sample of n animals were calculated, with n varying from 3 to 7 and values for x that were considered were n, n - 1 to n - x. This probability, among animals 9-24 months old, ranged from 0.70 to 0.96 for herds with PI animals and from 0.13 to 0.37 for herds without. Using the result of bulk milk testing in addition did not add to the prediction. It was concluded that, due to the high percentage and large variation of antibody positive animals in herds without PI animals, it is not possible to predict the presence of PI animal(s) in dairy herds in The Netherlands using these methods.