Over the years, increased traffic loads and material degradation due to weathering have caused damage to old masonry railway bridges and viaducts. Monitoring the response under traffic, including the evolution of damage, is paramount to understanding the complex serviceability behaviour. The current standard method for assessing damage in masonry bridges relies on visual inspections. Only severely damaged structures are monitored, considering medium-term changes in existing cracks and variations in vertical displacements at specific locations under train loading. This usually involves techniques that require physical contact with the structure and partially or fully occupying the monitored bridge spans. More advanced contactless methods, such as Digital Image Correlation (DIC), offer the potential for non-invasive monitoring. However, these optical-based techniques are primarily used in laboratory settings and are only rarely applied in the field. This paper examines the field application of DIC, comparing the effectiveness of 2D and 3D DIC techniques for monitoring real railway masonry bridges. 2D DIC can be directly applied for field monitoring with a simple setup. Still, it is primarily designed to capture in-plane displacements of planar surfaces with the adopted camera oriented perpendicular to the surface being monitored. In contrast, 3D DIC, although substantially more complex to set up, offers advantages for monitoring non-planar surfaces and the effects of out-of-plane movements. This is important for structures that exhibit non-negligible transverse displacements, such as masonry viaducts with tall piers. A detailed analysis of the results obtained using 2D and 3D DIC monitoring on real masonry viaducts under train loading indicates the potential and limitations of contactless monitoring via optical-based methods.