Foot-and-mouth disease (FMD) vaccine potency testing has historically been performed by experimentally infecting vaccinated cattle. A few alternative approaches to the in vivo challenge test based on the correlation between serum titres of primo-vaccinated cattle and protection against infection have been proposed, but none have been accepted by the European Pharmacopoeia (Ph.Eur.) due to the lack of statistical power and the pooling of data over time. The present study addresses these issues and presents data of 150 cattle vaccinated according to Ph.Eur. standards. Four laboratories took part in the serological testing and different serological assays were used, including virus neutralisation assays and ELISA formats. Models correlating specific anti-FMD virus antibody titres to protection were built using logistic regression followed by Receiver Operating Characteristic (ROC) analysis. The best models accurately predicted the in vivo protection status in 80.0% of the cases. Although differences were observed between laboratories and assays used, the majority of antibody pass-levels, determined using ROC analysis, corresponded to at least 75.0% probability of protection. The indirect potency assessment procedure proposed is at least as precise (repeatability=65.8%, reproducibility=60.7%) as the in vivo test, can be standardised and results in a quantitative PD50 value. The validity of the procedure was also demonstrated.