This paper addresses the problem of selecting Persistent Point Scatterers (PPS) in a stack of interferometric Synthetic Aperture Radar images suitable for accurate SAR parameters estimation. PPS are a subset of the well known Persistent Scatterers: besides the usual PS properties, they are characterized by an impulsive trend and a good degree of isolation from neighbor targets. Using their specific features, a simple and efficient algorithm for the PPS selection has been developed; a statistical model for PPS is also provided and experimentally validated. Results are illustrated using real X-band Cosmo SkyMed data.