Abstract: To address the problem of low classification accuracy in existing bioradar sleep staging technology caused by the use of indirect physiological features, research on bioradar sleep eye movement detection is conducted. A synchronous acquisition platform for millimeter-wave bioradar and electrooculogram (EOG)/polysomnography (PSG) is set up. Simulated eye movement experiments and real sleep experiments are carried out respectively. In the simulated eye movement experiment, the time-domain cross-correlation coefficient between the eye movement signals detected by the bioradar and the EOG signals exceeds 0.90. The frequency-domain main frequency error is less than 0.05 Hz. In the real sleep experiment, eye movement signals consistent with EOG are also detected by the bioradar. These signals exhibit explosive characteristics in the rapid eye movement period. They show intermittent smooth fluctuation characteristics in the light sleep period. They display stable low-amplitude fluctuation characteristics in the deep sleep period. The experimental results indicate that the bioradar can effectively detect and distinguish three typical sleep eye movement events: rapid eye movement, slow eye movement, and no significant eye movement. Thus, a new technical approach for bioradar-based sleep staging is provided. It is expected to further improve the staging accuracy.