The Shenzhou - 4 spaceborne (SZ - 4) altimeter waveforms were processed, and then the significant wave heights (SWH) was retrieved on the basis of waveform fitting and waveform retracking. Waveforms processing includes the waveform 1 s averaging, the elimination of thermal noise and the waveforms normalization. Double peaks were found on each SZ - 4 waveform, and it was pointed out that the region of waveforms with the second peak is abnormal and its effects on the whole waveform in the waveform fit should be taken into consideration. To obtain the width of the waveform leading-edge, a method was proposed to find the starting point of waveform, and the half-power point of waveform was found by retracking the waveform. The normalized waveforms were fitted with the Haynes model by using the weighting least square fit method. Then the selections of the weighting coefficients and their effects on significant wave hight retrieving were discussed, and the optimal five-region weighting method was proposed. At last, the SWH data of SZ - 4 altimeter retrieved by using the proposed method were compared with those of ERS - 2 and Jason - 1 altimeter, and it was concluded that the SZ- 4 altimeter can detect significant wave height.