Non-contact monitoring of oxygen saturation is required to diagnose obstructive sleep apnea, which leads to lethal sleep disorders in adults. This study proposes an estimation algorithm for oxygen saturation using a noncontact radar sensor. First, the algorithm estimates the displacement of the chest, respiratory rate, and cardiac rate using the phase fluctuations of complex signals received from an individual. Next, the pulmonary ventilation, tidal volume, and minute ventilation were estimated using a step-by-step process, leading to the prediction of pressures corresponding to carbon dioxide and oxygen. Finally, the oxygen saturation was calculated by a second-order transfer function using the predicted pressures. In particular, the proposed method can robustly estimate oxygen saturation despite clutter and unwanted body movements because of the use of phase fluctuations preprocessed by several filtering methods. Through simulations and experiments, we observed that the proposed method using a Doppler radar can robustly estimate the oxygen saturation of an individual with unwanted body movements, unlike conventional methods.