An extremely large amount of volcanic gas has been released since mid-August 2000 from the volcanic island of Miyakejima, Japan, after formation of a summit caldera of 1.6 km diameter. The volcanic gas emission was continuous with very little extrusive magma activity. Variation of the SO2 emission rate was monitored by repeated measurements with an airborne correlation spectrometer. In December 2000, the SO2 emission rate averaged for the month peaked at 54 kt/d, which is twice the global SO2 emission rate from nonerupting volcanoes evaluated before this activity. The SO2 emission rate gradually decreased, almost linearly when plotted on a log scale, to 7 kt/d by the end of 2002, and then remained constant until at least December 2003. The total SO2 emission amounts to 18 Mt, comparable to the emission of a large explosive eruption such as Pinatubo in 1991. A theoretical evaluation, based on the model of magma convection in a conduit, suggests that extremely large volcanic gas emissions can be caused by formation of a magma pathway with a slightly larger diameter than exists in common systems, because the magma-transport rate is proportional to the fourth power of the conduit radius. Because volcanic gas emissions were initiated by formation of a summit collapse caldera of 1.6 km diameter, the creation of a large magma-conduit system through fractures formed during caldera collapse is likely the underlying cause of the extremely large volcanic gas emissions from the volcano.