With more application of wide area measurement systems and fault recording networks to power systems, the electrical measurements can be obtained more convenient and reliable. This paper uses a comprehensive fault diagnosis method based on switching-status alarms and electrical measurements. By analyzing the switching-status data in alarm information and finding out the outage area, the suspicious fault element set can be determined. The relevant data is analyzed through fuzzy analytic model and Hilbert-Huang transform. The analytic fault degree、frequency distortion degree and energy variation degree are defined as indices of fault diagnosis. The three indices are fused as evidence forms by the improved D-S evidence theory to obtain the diagnosis results. Simulation and calculations show that the proposed method is effective of reducing the influence resulted from relay or breaker’s mal-operation or fail-operation. The presented approach may significantly improve the diagnostic accuracy of power grid faults.