The invention discloses a power distribution network fault handling method adopted under the condition of fault information loss or mistakes. The method includes the following steps that (1) fault-related remote signaling information is read, wherein after receiving fault signals, a master station of a power distribution automatic system executes a forced data call on related devices; (2) fault information is corrected, wherein a united fault information identification method is used, so that influences of distribution wire mistaken-reporting accident signals are removed; (3) faults are located, wherein the faults are located through an improved 'rough set' fault locating method; (4) the faults are isolated, wherein according to fault localization, a fault section is determined and is isolated through operation of a fault region boundary switch; (5) power supply is recovered, wherein power recovery of non-fault sections is achieved. According to the method, under the conditions that delay or loss or report omission or mistaken-reporting exists in the fault signals pushed by a power distribution network fault detection device, occurrence of the faults can be correctly judged, the faults are located and isolated, and power recovery of non-fault areas is achieved.