By extending our previous study by Maehara et al. (2012, Cat. J/other/Nat/485.478), we searched for superflares on G-type dwarfs (solar-type stars) using Kepler data for a longer period (500 days) than that (120 days) in our previous study. As a result, we found 1547 superflares on 279 G-type dwarfs, which is much more than the previous 365 superflares on 148 stars. Using these new data, we studied the statistical properties of the occurrence rate of superflares, and confirmed the previous results, i.e., the occurrence rate (dN/dE) of superflares versus flare energy (E) shows a power-law distribution with dN/dE{prop.to}E^-{alpha}^, where {alpha}~2. It is interesting that this distribution is roughly similar to that for solar flares. In the case of the Sun-like stars (with surface temperature 5600-6000K and slowly rotating with a period longer than 10 days), the occurrence rate of superflares with an energy of 10^34^-10^35^erg is once in 800-5000yr. We also studied long-term (500 days) stellar brightness variation of these superflare stars and found that in some G-type dwarfs the occurrence rate of superflares was extremely high, ~57 superflares in 500 days (i.e., once in 10 days). In the case of Sun-like stars, the most active stars show a frequency of one superflare (with 10^34^erg) in 100 days. There is evidence that these superflare stars have extremely large starspots with a size about 10 times larger than that of the largest sunspot. We argue that the physical origin of the extremely high occurrence rate of superflares in these stars may be attributed to the existence of extremely large starspots.