Based on 2 previous surveys on the occurrence of infection with bovine virus diarrhoea virus (BVDV) in Danish and Michigan dairy herds, the prevalence and incidence of the infection were compared. The presence of certain possible risk factors for the occurrence of infection in the 2 areas were summarized and it was investigated if any of these risk factors had significant effect on the presence of animals persistently infected (PI) with BVDV in the dairy herds. Information on the cattle population density in the 2 areas was obtained from statistical yearbooks. Further information for the individual farms on age distribution, housing of animals, herd size, pasturing and purchasing policy was gathered. The prevalence of PI animals was more than 10 times higher in Denmark as compared to Michigan. In herds without PI animals, the annual incidence of seroconversion as calculated from the age specific prevalence of antibody carriers varied in most age groups between 20-25% in Denmark and between 5-10% in Michigan. All investigated risk factors except for herd size were in favour of a lower prevalence of infection in Michigan. The use of having animals on pasture and at the same time having purchased more than 40 animals within recent 3 1/2-4 years were significantly associated with presence of PI animals in the dairy herds (p = 0.01) when tested by the Mantel-Haenszel chi 2. Using multivariable logistic regression, the occurrence of PI animals was found to be significantly related to the study area (Michigan and Denmark) as well as to herd size and purchase intensity.