The 2011 M9.0 off the Pacific Coast of Tohoku Earthquake (Tohoku-oki earthquake) brought a great impact on the long-term forecasting of subduction earthquakes around the Japanese islands. Short historic data together with dogmas of modern seismology, such as conventional asperity model, characteristic earthquake model, and earthquake scaling law prevented us to have anticipated the size of M9 earthquake offshore Pacific coast of Tohoku. It may suggest that longer than 1000-year earthquake occurrence history is required to properly evaluate the size and frequency of mega-thrust events, same as the M~7 destructive earthquakes associated with inland active faults. The Tohoku-oki earthquake has significantly changed the state of crustal stress in northeast Honshu island from EW compression to EW extension, in which numerous widespread triggered earthquakes have been occurring. Here I introduce the coseismic stress transfer due to the Tohoku-oki earthquake onto the major active faults, and then demonstrate the importance of the transient changes of state of stress on the faults for long-term earthquake forecasting during the next few decades.