DC electronic instrument transformers (DCEIT) are widely used in the voltage source converter based high voltage direct current transmission (VSC-HVDC) system to measure the bus voltage or current, and should be of accurate and quick transient response. The transient performance of the DCEIT need to be calibrated using a calibrator. The existing schemes are difficult to recognize different transient step responses automatically, and usually require human intervention, which limits their flexibility and applicability. Besides, the accuracy of the calculated results is affected by human factors. A selfadaptive algorithm of transient performance analysis is proposed. It recognizes the characteristics of the transient step response based on wavelet analysis, and distinguishes the steady-state and step intervals automatically. Then it evaluates and finds the stable data segments in the steady-state intervals adaptively to calculate the high and low steady-state values. Finally, the data in the step interval are interpolated to search the feature points of the waveform, and then the transient response parameters are calculated according to the coordinates of feature points. In the validation stage, six kinds of step response waveforms are generated by MATLAB to test the performance of the proposed algorithm. Compared with the boundary window method, the proposed algorithm could recognize the characteristics of the tested waveforms automatically and achieve a more accurate analyses of transient performance. In addition, a calibrator prototype based on the proposed algorithm is developed, and some tests are carried out at the National Center for High Voltage Measurement (NCHVM) in China, whose results verify the feasibility of the algorithm. 1