We demonstrate that it is possible to retrieve geometric and kinematic information about the source of a destructive past earthquake by inverting its regional macroseismic intensity patterns. In fact, in the case study the inversion results agree with the seismological instrumental measurements of the 1930s and with neotectonic evidence. This leads to the hope that more knowledge about preinstrumental events can be obtained: a key toward improving the calculation of seismic hazard, mostly in the Old World. After validating our technique on the 1987 Whittier Narrows, California, earthquake <Pettenati and Sirovich, 2003; Gentile et al., 2004> we achieve the present results by investigating an earthquake that happened in 1936 in northeast Italy. The automatic inversions were performed by using a simplified formula for body waves that radiate from a linear source. The inversion shows two minima on the hypersurface of the minimum residuals (calculated -- observed intensity at all sites) in the multiparameter model space. These two minimum variance source models resemble the two auxiliary planes of the same theoretical fault plane solution, similar to that given by the standard use of the first P wave arrivals. The present result encourages us to treat more cases and to explore new inversion techniques for quantitatively treating the intensity patterns of earthquakes because, at least in some cases, they carry geometric and kinematic information about their source.