A signal change in stator current often indicates that a variable-frequency motor is malfunctioning. In this study, we developed a method based on discrete fractional Fourier transform (DFrFT) to identify rotor defects in a three-phase induction motor. The first step was to measure the stator current in an induction motor, followed by the application of DFrFT to detect rotor faults. DFrFT is frequently used to transform a time-domain signal at different angles. Angles from 0–2π were divided into 20 equal sections, which were sequentially transformed to construct characteristic matrices. For clearer characteristic information, the fractal method was applied to extract features, fractal dimension, lacunarity, and the mean value from the pattern matrices. Finally, defect patterns were identified by applying extension theory. To verify whether the proposed method was feasible for rotor fault recognition in the presence of interference, ±5 to ±15% Gaussian white random noise was added to the current signal. The results indicated that the proposed method can diagnose various rotor defects in a motor.