The majority of existing buildings in the Mediterranean area suffers from structural deficiencies, since it was designed according to old seismic prescription or even considering gravity loads only. Such buildings are vulnerable to seismic excitation, do not meet the performance requirements imposed by current seismic codes and often show evident signs of decay. Hence they need to be retrofitted, but a proper selection of the seismic upgrading technique stems from a reliable assessment of their seismic response based on accurate numerical models. In this regard, nowadays it is widely accepted that the effect of non-structural elements should be considered to properly  reproduce the seismic response of buildings. For instance, the effect of infill panels should be taken into account by the numerical models that are developed for seismic assessment of structures. Indeed, several studies have demonstrated that infill panels significantly increase the lateral stiffness of the frame and influence its dissipative capacity and collapse mechanism. Unfortunately, despite all buildings are endowed with infill panels, the contribution of these non-structural elements to the structural response has been often “conservatively” neglected, in favour of computational simplicity.In this framework, this paper presents a technique for calibration of the finite element numerical model of r.c. frames with infill panels based on the results of a quasi-static cyclic experimental test of a prototype infilled frame. The proposed technique is based on a sub-assembly approach and relies of the fact that (1) the response of the system is mainly governed by the infill panel for small amplitude displacement, while (2) it becomes coincident with that of the bare r.c. frame for large amplitude displacements. Hence, the mechanical properties of the bare frame and those of the infill panel are separately extracted from the same cyclic response of the infilled frame. The sub-assembly approach allows the calibration of a numerical model “by layers” and leads to the advantage that the execution of two different experimental tests, one on the bare frame and the other on the infilled frame, is not needed, thus providing appealing cost and time savings. This technique has been here applied for the calibration of the numerical model of a one-bay one-storey r.c. frame with infill panel. A full scale cyclic test of the infilled frame is also executed and the obtained structural response is assumed as target to be fitted.