A retrospective analysis of seroconversion to Maedi-Visna virus (MVV) was carried out for 10 infected semi-intensively reared dairy-sheep flocks that were tested annually between 1994 and 1999. Four of the flocks raised replacement lambs artificially with bovine colostrum and milk replacement to avoid lactogenic MVV infection but did not prevent aerosol contact between replacements and other sheep in the flock. Flock culling percentages ranged between 14 and 25% and in eight flocks the number of sheep that seroconverted was similar to or lower than the number of sheep culled--suggesting that incidence could be reduced by culling seropositive sheep without increasing average culling percentages. Random-effects logistic regression indicated that seroconversion was associated positively with increasing contact with infected sheep and with lifetime MV-serological status of the dam (used as a proxy measure of genetic susceptibility), but not with mode of rearing pre-weaning (artificially or with a seropositive or seronegative dam). Our results indicate that when conditions allow efficient horizontal transmission, there is no evidence that lactogenic infection increases the risk of MV infection and that there is an important inheritable component of disease resistance or susceptibility.