Millimeter wave radar has been widely used in the field of vital sign signal detection, but the presence of respiratory harmonics often interferes with the effective extraction of heartbeat signals. Therefore, this article proposes a radar respiratory and heartbeat signal separation algorithm based on adaptive cancellation. Firstly, the human chest radar echoes are collected using millimeter wave radar to obtain all range bins containing targets; Secondly, phase extraction preprocessing is performed on different range bins, and signals on two range bins are selected based on the characteristics of micro motion signals on each range bin. These signals are used as the input signal and the expected signal for the adaptive filter, and the heartbeat signal is obtained through cancellation; Finally, the respiratory and heartbeat frequencies are obtained through autocorrelation. The radar test results show that the algorithm proposed in this article can effectively suppress respiratory harmonics in radar echoes.