Voltage sags caused by short circuit fault are one of the most severe power quality events. It is beneficial to the operation and management of a power system if the short circuit fault types can be identified. A short circuit fault classification method based on double dq transformation is proposed in this paper. The proposed method decomposes the measured three-phase voltage to positive and negative sequences first， and then transforms the two-sequence components into dq frames. Phase angle and root-mean-square value of the voltage are generated as characteristic values to identify the fault types. This method only requires a few logic and simple algebraic calculation， and also can be combined with voltage sag detection to form a unified voltage sag detection and fault classification method.