This paper described an experiment of diagnosing the mass and aerodyn amic unbalances of helicopter rotor based on fuselage vibrations. Both faults with different degrees were simulated on a helicopter rotor test rig, and the fuselage vibration signals were acquired. Spectrum analysis was made by using fast Fourier transform(FFT), and the relationship between amplitudes of 1／rev components of the fuselage vibrations and different degrees of the both faults wasillustrated. The faults were accurately classified by using a probabilistic neural network (PNN) and the fault degrees were identified by a radial-based-func tion network. The result shows it is feasible to diagnose the unbalance faults of a helicopter rotor by using fuselage vibrations without the rotor track information.