A synchronous generator is one of the most critical components of a power system. Accurate and timely fault detection of synchronous generators is necessary for the reliable and smooth operation of the power system. This chapter proposes an observer-based fault detection scheme for synchronous generators. The fault detection scheme is designed based on an adaptive neuro-fuzzy inference system. The salient feature of this scheme is that the fault detection system is independent of the analytical model of the generator, which is otherwise needed in observer-based designs. The proposed scheme incorporates the effects of nonlinearities and model uncertainties in its design. The IEEE 3-machine, 9-bus system is used to demonstrate the effectiveness of the proposed approach. It successfully detects faults like loss of excitation, step increase in mechanical torque, high-voltage bus voltage change, current transformer faults, potential transformer faults, and three-phase faults of synchronous generators.