Masonry arch bridges, mainly built over a century ago, still represent a significant part of the railway and road infrastructure in Italy and Europe. The long-term structural stability and integrity of these structures, in addition to factors such as lack of maintenance, material degradation, and the increase in axle loads and traffic volume, are also challenged by horizontal loads, such as seismic actions. In this context, the National Seismic Observatory of Structures in Italy started a program to monitor existing masonry bridges, allowing the detection of oscillations caused by earthquakes and providing crucial data to better understand the seismic behavior of these structures. This paper presents the dynamic identification of a monitored multi-span masonry arch bridge located in Southern Italy. The bridge is modeled using a three-dimensional discrete element approach, representing it as an assembly of rigid and deformable blocks that include both structural and non-structural components, such as piers, abutments, arch vaults, backing, spandrel walls, and filling material. The bridge geometric properties and material properties are defined based on in-situ survey data. The fundamental modes of vibration of the masonry arch bridge are determined, not directly through eigenvalue analysis, but via dynamic identification using appropriate input motions. The resulting mode shapes and periods are compared with those obtained from structural monitoring. The influence of different boundary conditions for the piers and abutments is analyzed to assess their impact on the results. Finally, sensitivity analyses are performed to assess how variations in the model details, such as the type and number of employed blocks, affect the outcome accuracy and computational efficiency.