Isotype-capture ELISAs for BCV-specific IgA and IgM were developed and tested on milk and serum samples from Swedish cattle. The capture ELISAs showed higher sensitivity than indirect ELISAs for detection of BCV-specific IgA and IgM. In the capture ELISAs the agreement between detection in milk and serum samples was 94% for IgA and 86% for IgM. The correlation between log(10) titres in milk and serum was r=0.82 (P<0.001) for IgA and 0.84 (P<0.001) for IgM. Milk seemed a better target than serum for diagnosing specific IgA at low levels. There was no variation in the isotype-specific BCV antibody titres between healthy quarters of the same udder, but subclinical mastitis was associated with higher levels of IgA antibodies and weak false IgM positive reactions in undiluted milk. Bovine IgA and IgM antibodies in milk and serum showed high stability towards freezing and thawing and storage at room temperature. The antibody responses to BCV were followed in milk and serum from six dairy cows and in serum from four calves for a period of 1 year after an outbreak of winter dysentery (WD). In this outbreak some animals became reinfected with BCV. The IgA and IgM capture ELISAs differentiated between primarily BCV infected and reinfected animals. In the primarily infected cattle, IgM antibodies were first detected in milk and serum four to nine days after the first WD symptoms observed, and were subsequently detected for at least 2-3 weeks. IgM was also detected in the reinfected cows, but mostly at lower levels and for a shorter period of time than in the primarily infected animals. In milk, however, the IgM response of the reinfected cows was detected for a longer period of time than in serum. Six months after the outbreak, IgA was still detected in both serum and milk of all six cows and also in serum of one calf. The reinfected cows showed higher and more long-lasting peak levels of IgA in milk and serum than the primarily infected cows, indicating boosting of the IgA response.