As the conventional FFT-based extraction algorithm of vital signals for the through-wall detection radar（TWDR） is insufficient in handling non-stationary signals and is sensitive to respiration harmonic interference,a novel method to extract vital signals is proposed based on the empirical mode decomposition（EMD）.In this method,the vital signal accepted by the radar is decomposed into several intrinsic mode functions（IMFs） via the EMD,and the IMF components describing the structural characteristics of the vital signal are used to reconstruct respiration and heartbeat signals in the time domain.Simulated results indicate that the proposed method helps to extract heartbeat signals with high accuracy because it effectively eliminates the interference of respiration harmonics.