Abstract: Noise reduction for interferometric phase images in synthetic aperture sonar(SAS) is addressed in this paper.The phase noise is modeled as additive noise.Four filtering algo-rithm are analyzed.They are pixel-averaging,conditional neighborhood-averaging,morphological and modified median filtering,and vector-filtering.Using the interferometric phase images generated with simulated data,it is found that only the morphological and modified median filtering algorithm and the vector-filtering algorithm can produce accurate height maps.By comparison of residues and the height map variance,it is concluded that the vector-filtering algorithm has the best performance.