A specific reverse transcription polymerase chain reaction (RT-PCR) for the detection of the polymerase gene (3D) of foot-and-mouth disease virus (FMDV) was developed and validated with an analytical sensitivity of equal to, to 1,000 times higher than that of a single passage virus isolation. The performance of the RT-PCR was determined in 180 runs. After implementation, 5.3% of the tests had to be rejected due to invalid controls (e.g. cross-contamination of negative controls). The diagnostic sensitivity, determined using 124 samples from experimentally infected animals, was 91.9% for RT-PCR and 84.7% for virus isolation. Diagnostic specificity, determined by testing 258 samples from uninfected animals, was 100% by both tests. Of the 627 samples tested by RT-PCR and virus isolation, 85 reacted positively in both tests (13.5%) and 447 negatively in both tests (71.3%). One sample was positive by virus isolation and negative by RT-PCR (0.2%), 94 samples were positive by RT-PCR and negative by virus isolation (15%). The majority (84 of 94) of the 15% RT-PCR positive and virus isolation negative samples were among other samples from farms that reacted positively by both tests. The new RT-PCR is a robust, reliable and sensitive test, provided that adequate measures are taken to prevent cross-contamination. A possible preventive measure is to exclude ELISA positive samples from the RT-PCR testing.