In this paper, a new approach that combines Chi-Square Test and Fuzzy ARTMAP (FAM) neural network mapping is proposed to Fault Diagnosis of Integrated Navigation Systems. Stale Chi-Square Test on whole state vector and individual state is performed to detect faults. Then, the pattern of State Chi-Square Test result is used to isolate and identify faults by using two separated FAM networks respectively. The fault magnitude and fault occurring time can be identified by the second FAM.