The invention provides a method for identifying a state of a power transformer winding by using a pseudorandom sequence. The method comprises the following steps: adopting an M sequence in the pseudorandom sequence as a signal source; after exciting a winding system, collecting input and output signals by using double-channel digital collection equipment; calculating to obtain a pulse response transferring function of the winding system according to the input and output signals; then carrying out correlation function calculation on the input and output signals to obtain a pulse response of the winding system; then describing the state of the winding according to a derivation parameter and a frequency response curve; and judging whether the transformer winding is deformed or not. The testing method disclosed by the invention can be used for comparing amplitude-frequency response curves of the transformer winding and has the consistency. The method disclosed by the invention has the characteristics that the pulse response transferring function of the winding can be directly obtained, the measurement time is short and the like; and furthermore, the method disclosed by the invention can judge the state of the transformer winding according to the change of specific parameters, including poles, null points and the like, of the transferring function, so that the problem in the prior art that technicists easily carry out erroneous judgment is solved.