To eliminate porcine reproductive and respiratory syndrome virus (prrsv) from a supplying boar stud, samples of serum and semen from 118 boars were assessed three times a month by an indirect fluorescent antibody (ifa) test to detect antibodies, and by a nested reverse transcriptase-pcr (nrt-pcr) to detect the genome of prrsv. The boars detected as persistently infected carriers were culled. A prrsv-negative population of boars was established after three months and no semen positive for the virus was detected for six months. Subsequently, a three-step plan was introduced to eliminate prrsv from the seedstock breeding farm during three parity cycles on the farm over 15 months, each step taking five months. In step 1, umbilical cords taken from piglets at birth and serum samples taken from their dams at the start of weaning were subjected to ifa and nrt-pcr analysis. The sows with prrsv detected in serum by nrt-pcr were regarded as carrier sows and culled. The rates of detection of prrsv were reduced from 5 per cent to 2·5 per cent in the sera of the sows, and from 14·8 per cent to 1·8 per cent in the umbilical cords of the piglets. In step 2, the sows that had farrowed the piglets with prrsv detected by nrt-pcr in their cords were considered to have transmitted the infection and removed. During step 2, the virus detection rates in umbilical cords by nrt-pcr were reduced, but not completely eliminated. In step 3, 10-week-old nursery pigs with antibodies to prrsv in their serum by ifa and elisa were culled. The three steps established the prrsv-negative state of the multisite farm containing the breeding and nursery farm, and the prrsv-negative state of both the multisite farm and the supplying boar stud was evaluated by monthly monitoring over at least one parity cycle of the farm for five months.