In this paper, a Novel approach for diagnosing gearbox faults in transient speeds using the Motor Current Signature Analysis (MCSA) is introduced. Current and speed signals were collected from the motor in ramp-up speeds under no load. For increasing the signal-to-noise ratio, a dual-level time synchronous averaging is introduced. First, the sinusoid component related to the motor supply frequency is removed from the signals. Then, the tachometer signal is used to perform time synchronous averaging on the residual signal. Features extracted in both angle and order domains clearly distinguished the gear conditions of normal, pitting and eccentricity.