Though radar image interferometry is a relatively recent technique, it is largely used by the remote sensing community because it is cheaper than most of the traditional methods of relief and surface deformation analysis. Interferometry uses the phase information of radar images to measure distance variations. The analysis of phase difference between two radar images allows the construction of Digital Terrain Models (DTM) with meter accuracy. Having a DTM or a third image, ground movement between images acquisition can also be measured with millimetric accuracy. The performance of this technique is limited by temporal and geometrical decorrelation and atmospheric inhomogeneities. These perturbations can be corrected using coherent pixels over long time intervals (called Permanent Scatterers). They can be calculated over large series of SAR images and be used as a GPS network to monitor ground mouvement caused by earthquakes, seismic faults, volcanos and urban subsidence. Due to their temporal stability they can be used to model atmospheric deformation and to achieve atmospheric effects remotion on SAR interferograms [2]. At this time, many different methods have been proposed to extract “stable” pixels and some of them are based on very different principles (amplitude or coherence or phase analysis criterion). The aim of this paper is to study the link between two apparently different identification methods by means of a theoretical and a practical analysis. The first section recalls the principle of both methods underlining their differences for the stable scatterer selection while the second section compares the extracted pixels on a region of interest, the Gulf of Corinth and analyses the relationship between the two sets of pixels. The first method is the well-known method proposed by Ferretti et al. [1]. Although the pixel characterization is their phase stability, the pixel selection is based on image amplitude values. The dispersion index (temporal variation coefficient) is thresholded which means that assuming a Rice distribution of the pixel amplitude, phase standard deviation is low. In the second method, proposed by Chaabane et al. [2], stable scatterer selection is based on interferogram information and a tropospheric model assuming an horizontally homogeneous troposphere and a linear correlation between phase and elevation of stable scatterers. Both methods have been used on real images of the Gulf of Corinth to select the stable scatterers. The set of pixels is then divided in three sub-sets: common pixels and pixels detected by only one of the methods. They are then characterized depending on the ground nature of the pixels (rocks, urban areas, ...) and conclusions about both selection methods are derived. References [1] A. Ferretti, C. Prati, F. Rocca, Permanent Scatterers in SAR Interferometry, IEEE TGARS 2001, vol.39, 8-20. [2] F. Chaabane, A. Avallone, F. Tupin, P. Briole, H. Maître, Correction of local and global tropospheric effects on differential SAR interferograms for the study of earthquake phenomena, IGARSS, jul 2003, Toulouse, France.