Analysis of temporal sequences of satellite images is of great importance in the monitoring of environmental phenomena, where both multi-temporal and multi-spectral images are widely used. The first problem to deal with such a kind of imagery is their geo-referencing, i.e. computing a transformation to a geodetic or cartographic datum. This task is usually performed by manually measuring control points, whose ground coordinates are read on existing maps or measured by GPS. In this paper we propose a software (GEOREF) to increase the automation of this procedure, which is very heavy for the operators in the Remote Sensing field. Considering a multi-temporal sequence of satellite images, instead of geo-referencing all the images by interactive measurement of ground control points, you have to manually register only one image of the sequence to the geodetic datum. Then control points can be automatically extracted in the other images, which can be registered to the first one and thus to the ground. The paper would like to give an overview of the algorithms involved in the proposed procedure and its implementation. Furthermore some applications of GEOREF to register Landsat TM and ETM+ images are presented.