Scour is the main cause of bridge failure in the United States and around the world. It involves the loss of soil from around bridge foundations due to fast flowing water. Monitoring bridge scour is challenging as it occurs underwater. A scour monitoring technique that uses accelerometers on the bridge piers and deck to provide real-time measurements of ambient vibration, from which scour-induced changes in the natural frequencies of the bridge may be monitored, has recently emerged. However, field validation of this method is difficult as scour is hard to predict -a monitoring system would need to be installed on a candidate bridge with no guarantee that natural scouring would occur within the trial monitoring period. Previous research on vibration-based scour monitoring has therefore been limited to laboratory-based experiments and numerical simulations. This paper presents a field study in which a bridge with preexisting scour was monitored throughout a repair program, which involved the controlled raising of the riverbed level (i.e. scour in reverse). Ambient vibration due to vehicle traffic over the bridge was monitored before, during and after the repair, using eight high-sensitivity accelerometers and a remote data acquisition system. The riverbed profile was measured with sonar scanning before and after the repair. Numerical modeling predicted a 3-5% change in the first two natural frequencies of the bridge due to the scour repair. However, the actual frequencies measured on site, derived using operational modal analysis (in this case, Frequency Domain Decomposition), showed a high degree of variation. Alternative structural response parameters (mean power spectral density at the modal peaks and the modal amplitude) were found to capture the change in behavior resulting from scour. The opportunities for using these alternative parameters as a measure of scour, and the challenges involved in using operational modal analysis for scour monitoring on a real bridge, are discussed in detail.