For pilot column high-frequency dithering in civil aircraft autopilot system, a signal processing method based on wavelet analysis and finite impulse response (FIR) is proposed to reduce pilot column high-frequency dithering. Firstly the principle and effect of column high-frequency dithering is discussed, and wavelet theory-based toolbox is used to discompose input signals of control law in autopilot system, by which the frequency in signal can be determined. Finally FIR filter is designed to eliminate the unuseful frequency part, and then reduce the output control signal of autopilot system, thus alleviating the column high-frequency dithering.