To provide the network management system that can specify a problem effectively regardless of a small use amount of memory capacity.

The system includes a management object(MO) class state pattern list management section 70 that manages a MO class state pattern list 72 defining a status pattern of each MO class based on a repository storing MO configuration information, a global event definition, and in-MO-class and inter-class event propagation definition, a Problem-MO status correlation management section 80 that manages a Problem-MO status correlation table 82 where a correlation between a problem event and an MO class status pattern receiving the effect of the event is described by referring to the MO class status pattern, and deduction sections 78, 84 that deduce a cause to the fault based on the Problem-MO status correlation table 82 and the MO class state pattern list 72 in response to fault event information 76.