The invention discloses an identification method for excitation surge current of a transformer based on kurtosis and skewness. The identification method comprises the following steps: initially setting a first setting value, a second setting value and a third setting value, performing uniform continuous sampling on differential current of the transformer by adopting an assigned sampling frequency to obtain a sampling value sequence; if a current effective value of the differential current is smaller than the first setting value, exiting the procedure, and otherwise, aiming at the current sampling value sequence, calculating the kurtosis and the skewness of the differential current of the latest power frequency cycle; if the kurtosis is greater than the second setting value and the skewness is greater than the third setting value, judging that the transformer is in an operation state of the excitation surge current, and locking differential protection of the transformer, otherwise, judging that the transformer is in other non-excitation surge current operation states; returning to continuously process the current effective value of the differential current. The identification method disclosed by the invention has the advantages of obvious criterion distinguishing degree, high correctness of identification result, low requirement on hardware realization, short judgment time, good real-time property and low requirement on hardware of an implementation device.