In order to carry out the research on the statistical distribution of power grid fault and its mechanism, the fault data of Nanchang distribution network from 2006 to 2009 are investigated in the article. It is discovered that the complementary cumulative distribution function (CCDF)of fault appears as a line on the log-log plot, which indicate it follow power law distribution. Whereas, the CCDF of truncated fault appears as a line on the log-linear plot, which indicate it follows exponential distribution. Moreover, detrended fluctuation analysis show that the fault data behaving long term autocorrelation. Records of the days with largest number of fault and meteorological data suggest that the extreme climate contribute to the power law distribution of fault in Nanchang distribution network.