Abstract: ‍ ‍A multi-life signal detection method based on variational mode decomposition （VMD） for wideband chaotic radar was proposed and experimentally demonstrated. Firstly， the original echo matrix was obtained by the chaotic correlation ranging. Then， the noise and clutter were suppressed and the multi-life signals were reconstructed based on the VMD method， which includes reducing the VMD computation burden by the decision with 0 dB peak noise ratio （PNR）， optimizing the K value through the central frequency analysis， and automatically extracting the respiratory signal by the 3 dB peak sidelobe level （PSL） decision. Finally， combined with the fast Fourier transform （FFT） and the constant false alarm rate （CFAR）， the respiratory frequencies and ranges of human targets behind the wall were simultaneously estimated. The experimental results show that the proposed method can accurately and quickly detect two to five human targets behind the wall with 20 cm thickness， and has a high range resolution of 15 cm.