Rotor faults in a synchronous generator severely deteriorate its performance, reducing the operation efficiency and reliability. Most of the field-wide accepted methods require stoppage of generator operation and disassembly, and online diagnosis of rotor faults is difficult due to the lack of dependable online methods for sensitive rotor fault detection. Therefore, online methods that can reliably and unambiguously detect rotor field winding and eccentricity faults before they lead to machine failure are required. This paper presents a new method for detecting and distinguishing rotor field winding insulation and eccentricity faults online during synchronous generator operation based on measurements from airgap flux probes, which are installed in most synchronous generators used for electric power generation. A non-complex, time domain analysis method is proposed which makes use of the voltage induced in the airgap flux sensor for a reliable prediction of the presence and identification of rotor field winding turn insulation failure and rotor dynamic eccentricity. The method is tested and verified experimentally on a 5 hp synchronous generator and on measurements from a 330 MW pumped storage generator.