A longitudinal study was performed in a Swedish dairy herd where Neospora caninum had been isolated from a stillborn calf. Starting in autumn 1994, blood samples from all female animals in the herd were collected once yearly until 1999. The sera were analysed for presence of IgG1 antibodies to N. caninum by the iscom ELISA, and by an avidity ELISA to establish the timing of infection. In addition, data on reproductive performance were compiled. During the study the percentage of seropositive female animals increased from 63% to 87%. In 1994 a large number of young animals tested seropositive although their dams were seronegative, indicating that a transmission of the parasite other than the vertical had recently occurred. Low avidity values supported this assumption. The annual abortion rate increased from a mean of 2% before the initiation of the study to 9% in 1994-1998. During the same time, as judged by the avidity data, a large proportion of the animals shifted from being recently to being chronically infected. The source of the external infection in the herd could not be identified.