In the processing of data acquired by conventional radar altimeters, algorithms designed to compute elevation of signal backscatterers assume that the point closest to the satellite antenna is the nadir point. While this is a valid assumption over the oceans, it is often not valid in non-oceanic surfaces. Radar altimetry data acquired over ice sheets, for instance, are dominated by multiple scatterers off-nadir to the satellite, which implies that more than one scattering surfaces in the radar footprint contribute to the received signal. A SAR/Interferometric Radar Altimeter, which is the primary payload of CryoSat, allows the geolocation of an off-nadir scatterer by recording the receiving signal in two distinct receiving antennae separated by a baseline of 1.2 m. In recent studies on ice sheet elevation changes based on CryoSat data, however, only one scatterer is ‘tracked’. Hence, the full potential of the backscattered signal is not exploited. In this study, we design an algorithm that enables elevation estimation and geolocation of multiple off-nadir scatterers within the radar footprint from CryoSat Level 1b waveforms. With the results obtained by our CryoSat level 2 data processor, we obtain elevation change measurements over a part of the Jakobshavn Isbræ (or the Jakobshavn drainage basin) at West-Greenland. Our algorithm provides about twice as much retracked elevations compared to the number of elevations from the CryoSat ESA level 2 products, and hence a better spatial sampling of the elevation change signal.