We investigate the rupture process of Hyuga-nada earthquake of April 1, 1968 (M_JMA 7.5). Applying a multiple-time window inversion scheme to teleseismic body wave data, we obtained a detailed spatio-temporal distribution of moment release. The main source parameters are: the seismic moment=2.5×10²⁰[Nm]; the rupture area=64×48[km²]; the stress drop=3.4[MPa]; the focal depth=15[km]. The rupture consists of three major asperities: the first asperity centring about 10km south and 20km west from the hypocenter and having a maximum slip of 4.0m, the second one centring about 8km north and 5km east from the hypocenter and having a maximum slip of 3.0m, and the third one centring about 50km west from the hypocenter and having a maximum slip of 3.2m. We compared the rupture area with that of a few large events (M>6.5) subsequent to the 1968 event. Then we found that the above three asperities of 1968 event coincide with the low seismicity area in the Hyuga-nada region, and do not overlap with the source area of the subsequent large events (M>6.5). This rupture pattern and the seismicity suggest that an area of slab bending as well as fracture of the slab can behave as barriers during earthquake rupture. These barriers may control the maximum size of earthquake source in this region.