Abstract-Condition monitoring and fault diagnosis (CMFD) of drivetrain gearboxes has become a prominent challenge in assorted industries. Current-based diagnostic techniques have significant advantages over traditional vibration-based techniques in terms of accessibility, cost, implementation and reliability. This paper proposes a current-based, frequency demodulation-aided CMFD method for drivetrain gearboxes. A mathematical model is developed for a drivetrain consisting of a two-stage gearbox and a permanent magnet synchronous generator (PMSG), from which the characteristic frequencies of gearbox faults in the PMSG stator current are derived. An adaptive signal resampling method is proposed to convert the variable fault characteristic frequencies to constant values for the drivetrain running at variable speed conditions. A demodulation method, combining the Hilbert transform, a finite impulse response (FIR) differentiator, and a phase unwrapping algorithm, is developed to extract the instantaneous frequency (IF) patterns that are related to the faultinduced gearbox vibration. A fault detector is proposed for diagnosis of gearbox faults using statistical analysis on the extracted fault signatures. Experimental studies are carried out to validate the effectiveness of the proposed method.