Historically, the most common cause of bridge failure has been “scour” – the gradual erosion of soil around bridge foundations due to rapid water flow. A reliable technique to monitor scour could potentially guide timely repair and, in turn, mitigate the risk of future scour-induced bridge failure. Currently, there are various, mostly underwater, techniques employed by bridge managers to monitor scour, ranging from diving inspections to autonomous underwater vehicles; however, none have gained wide acceptance. A particular disadvantage of underwater monitoring techniques is that the equipment underwater is relatively difficult to install and prone to damage from fast-flowing water and debris. One possible solution might be to use a vibration-based method to monitor scour indirectly, using changes in dynamic modal parameters (e.g. the natural frequency of vibration) captured by sensors mounted on the bridge deck or piers above the water level. There has been extensive research into the use of vibration-based monitoring methods to identify other causes of failure, such as cracking and deterioration in bridge superstructures; however, this has proven to be ineffective in practice, as the expected sensitivities in modal parameters were only single-digit percentages and therefore insufficient to overcome environmental/operational sensitivity. In contrast to superstructure damage, scour is a special damage case, which changes a boundary condition of the bridge in the form of an increase in effective pier height as a result of the lowering of the ground level and therefore, significant changes in modal parameters can be expected. Recently, this concept has been studied primarily using numerical modelling simulations of a hypothetical integral bridge with piled foundations. Only one modal parameter – natural frequency – was investigated in most of these studies and it was predicted to change by up to double-digit percentages due to scour. Although such a high change could potentially overcome environmental and operational sensitivities, a critical problem is that this has been difficult to observe in practice with experiments on either real field bridges or small-scale soil-structure models. Another problem is that there is little knowledge of the applicability of this technique to different types of bridges and forms of scour. This research proposes a vibration-based technique based on a combination of three vibration parameters (spectral density, mode shape and natural frequency), which were studied using first-of-a-kind experiments and numerical modelling simulations on various types of bridges and forms of scour. A field trial was carried out on a bridge with pre-existing scour, which was monitored for ambient vibrations throughout a repair process involving controlled scour backfilling, i.e., “scour in reverse”. The effect of this scour backfilling was captured by measuring changes in two of these parameters, mode shape and spectral density, derived from the ambient vibrations. The mode shapes, in particular, showed the potential to localise the presence of scour to a specific pier. The most commonly measured vibration parameter of natural frequency was also observed from ambient vibrations, but this did not capture the effects of backfilling due to high measurement uncertainties. In order to study all three of these vibration parameters in a controlled environment, a centrifuge model testing programme was developed. These tests considered small-scale models representing three full-scale bridges with different bridge deck and foundation configurations (i.e. integral/ simply supported decks and shallow/deep foundations) and two forms of scour (i.e. local/global). The observed results of these small-scale centrifuge models were used to calibrate numerical models of full-scale bridges representative of these centrifuge models. Numerical simulation techniques were also developed to simulate the experimentally observed effects of local and global scour. These centrifuge experiments and the associated numerical modelling found that vibration-based methods have broad applicability for bridges, although only some parameters showed sufficient sensitivity to be viable as a monitoring technique in certain types of bridges. For example, the centrifuge bridge models with a shallow foundation did not show a significant change in natural frequency or mode shapes, but they did show a significant change in modal spectral density. This research therefore concludes that a vibration-based scour monitoring technique, examining the combined effect of natural frequency, mode shape and spectral density parameters, has significant potential to measure and even localise the change of scour depths at bridge foundations.