This paper proposes and compares five methods for phase estimation to measure slight change of chest movement with respiration using a dual type microwave Doppler sensor. A body direction to the sensor affects the performance of the respiration measurement, because microwave reflection is sensitive to the surface direction. The phase estimation from two sensor signals is the most important part to measure respiration. Thus, we developed new five methods for phase estimation. These methods were evaluated by calculating correlation coefficients between estimated waveforms and reference ones. The results demonstrated that the phase estimation based on least square method is the best for respiration measurement with respect to both waveform estimation accuracy and calculation time.