On October 2002, a seismic swarm occurred on Mt. Etna. One of the strongest events caused severe damage, up to a European Macroseismic Scale intensity of VIII that contrasts with its local magnitude of 4.4. The occurrence of significant damage at such a small magnitude is repeatedly observed in the area and is traditionally attributed to shallow source. Recorded strong-motion accelerograms and broadband seismograms demonstrate that there is one more cause for the severe damage, that is, an anomalously strong low-frequency (0:1 < f < 1 Hz) radiation deviating from the conventional Brune (1970) spectral scaling. Therefore, these earthquakes cause large ground displacements and long (≈20 sec) durations of shaking. The integration of digital accelerograms yields a maximum peak ground displacement as large as 1.8 cm at a distance of 18 km. Based on the sharp local attenuation of ground motion in the study area, we infer that peak ground displacements near the epicenters did exceed 10 cm. The occurrence of large displacements caused selective damage to medium-rise (≥ 3 stories) reinforced concrete buildings and elements like church façades.The frequency cutoff below 1.25 Hz in the Wood-Anderson response attenuates the peak-to-peak amplitudes used to assess local magnitudes. Therefore, M L values are not representative of the real strength of volcanic earthquakes. Because a prompt magnitude (and damage potential) assessment is crucial for civil protection actions, a procedure is proposed which, in near-real time, can be successful in identifying potentially damaging earthquakes of Mt. Etna through the computation of pseudovelocity response spectra. The procedure provides a magnitude value that is derived on a statistical basis from the Housner (1952) spectral intensity computed in the low-frequency band. This parameter is a suitable near-real-time indicator of large earthquake-induced building shaking and could also be applied for a preliminary estimate of the epicentral macroseismic intensity.