The high seismic hazard of the Italian territory and the vulnerability of its historic masonry heritage require the development of fragility curves that must be increasingly reliable and robustly correlated to exposure. To date, national-scale seismic risk analyses mainly use empirical curves derived from the statistical analysis of damage induced by past events. These curves have shown good reliability, but they correlate only with a few typological-structural characteristics of the building, such as the number of floors, the vertical structure typology or the construction period. The present research paper aims to overcome this limitation with a hybrid approach that provides a better exposure characterisation. Specifically, the proposed strategy integrates the SAVE and Piecewise Rigid Displacement (PRD) methods. SAVE is an empirical approach based on the damage assessment due to past seismic events used to identify a seismic behaviour of a structure, while the PRD method is a numerical approach that solves the boundary value problem for normal, rigid, no-tension material. It can model different structural typologies, and as a result, it also provides the value of the horizontal static multiplier that drives the masonry construction to collapse. An extended numerical campaign is carried out considering a sample of 750 masonry buildings distributed throughout the Italian territory and extracted from the PLINIVS typological database. Looking at each construction, first, a PRD analysis is conducted to define its seismic capacity, paying special attention to modelling construction details. After that, the SAVE method is used to classify the construction in a specific seismic vulnerability class, i.e., from A to C, with decreasing vulnerability. All the buildings belonging to the same class are then collected, and three fragility curves representative of the collapse state (one for each vulnerability class) are derived and validated against empirical and analytical ones commonly adopted in the Literature. The integrated methodology shows a good agreement between simulations and observations, confirming the viability of the proposed hybrid methodology for the large-scale assessment of masonry buildings, providing an effective strategy to plan mitigation and rehabilitation interventions.