Abstract: In order to suppress the random signal fluctuations caused by the echo noise of Mie scattering atmospheric probe laser radar, a new adaptive segmentation smoothing method was designed to smooth the signal. According to the characteristics of lidar signal, the effective changing amplitude of the signal is greater than the change of background noise. The amplitude of the signal noise could be expressed by several times of the standard deviation of the background noise. After comparing the amplitude difference of the adjacent signal with the noise amplitude, the effective position of signal change was determined. These positions could be recorded as the segmental endpoints of the signal. The common moving average algorithm was applied to each segment of the lidar signal. The method was examined by using the actual micro pulse lidar signal and compared by the common fixed segment algorithm. The results show that the adaptive segment smoothing method can choose the size of the smoothing window automatically according the intensity of the signal change. The noise is suppressed effectively and the signal distortion caused by excessive smooth is avoided.