In the 1940's, several destructive earthquakes occurred in western Japan. Seismograms in this period were usually recorded on smoked paper and the quality was poor compared to modern digital data. But the recent development of image processing technology enabled us to reconstruct feasible waveform data, whereby we investigated source rupture processes of two devastating earthquakes: the Tonankai earthquake (M7.9) of December 7, 1944, and the Mikawa earthquake (M6.8) of January 13, 1945. The results for the Tonankai earthquake show that the source roughly consists of a single asperity with a length scale of 100 km, having no segment structure with a smaller length-scale. Such a feature seems to be reflected to the sea bottom topography above the source region. The main source parameters are as follows: the seismic moment = 1.0 × 10 21 Nm (Mw=7.9); the fault area = 140 km × 80 km; (strike, dip, rake) = (225• , 15• , 79 • ); the maximum and averaged dislocations = 4.4 m and 3.0 m, respectively. The analysis of the seismograms for the Mikawa earthquake shows that the source is a reverse fault with a slight left-lateral component. The pressure axis is directed to ENE-WSW, which is a little rotated from the EW compression axis prevailing in western Japan. This fault can be regarded as the southern extension of the Nobi earthquake fault system. The main source parameters are as follows: the seismic moment = 1.0 × 10 19 Nm (Mw=6.6); the fault area = 20 km × 15 km; (strike, dip, rake) = (135• , 30• , 65 • ); the maximum and averaged dislocations are 2.1 m and 1.1 m, respectively. The slip distribution mainly consists of two asperities: the one near the hypocenter and the other 10-15 km northwest from it. The heavily damaged area is well correlated with the northwestern asperity.