A cross-sectional study was made in a major cattle-breeding area of Switzerland to investigate the prevalence and local distribution of animals antibody positive to bovine viral diarrhoea (BVD) virus and of persistently infected animals. The sample size and statistical analysis took into account the possible clustering of persistently infected animals on individual farms. Of 3440 animals tested on 121 farms, 1982 were found to have antibodies to BVD virus (95 per cent confidence interval for the population prevalence: 57.6+/-4.5 per cent) and 22 were persistently infected (estimate for the population prevalence: 0.64+/-0.34 per cent). The detection of persistent infection in animals less than seven months old was improved either by retesting one year later, by using an antigen-capture ELISA or, in selected cases, by the analysis of original serum samples by reverse-transcription PCR. The results showed that the prevalence of persistent infection may be underestimated by as much as one-third when antigen detection alone is used. None of the 121 farms investigated was free of antibody-positive animals and one in eight herds had at least one animal that was persistently infected. The prevalence of antibodies was lowest in seven- to 12-month-old animals (22.9 per cent) and highest in animals aged five years or over (84.9 per cent).