Macroseismic observations are a valuable tool, that can be used for the verification of site response and attenuation. Hence, in the framework of a bilateral project supported by Central European Investment Fund and with the extensive support of Global Earthquake Model Foundation to update the national seismic hazard model of Albania, intensity maps that contained more than 6000 macroseismic observations for 168 shallow (h<40 km) medium up to large (3.74-6.94) earthquakes that occurred in Albania and the surrounding area between 1851 and 1990 were digitized and analyzed. The Boxer methodology was used to redefine the location and magnitude of these events. An intensity attenuation model was calibrated for Albania by means of the same set of calibrating events, i.e. earthquakes of known instrumental magnitude and with reliable MDP distributions. In addition, a linear conversion relation from epicentral intensity I0 to Mw, and a new relationship between the expected intensity at the epicenter (IE) and Mw were also derived, from the same dataset. Instrumental magnitudes used in the calibration included both moment magnitudes derived from moment tensor solutions, and proxy values in order to ensure both a magnitude (or intensity) and temporal coverage as wide as possible. As intensity is a qualitative measurement of ground motion, several attempts have also been made in previous studies to derive some more quantitative relationships between surface geology and local amplification, by proposing also relations between the average horizontal spectral amplification and the intensity increment, and peak ground motion values (PGV) and macroseismic intensity. Hence, by taking into account these regressions applied on our macroseismic database, this study aims to evaluate whether the analyzed macroseismic data show good agreement with the expected seismic response, referring to soil type categorization on a national scale and the Ground Motion Prediction Equations selected in previous studies.