The Italian Lampedusa island in the Mediterranean Sea was chosen by the Centre National d'Etudes Spatiales (CNES) to support the verification of altimeters on board the TOPEX/POSEIDON satellite. The orbit was phased in such a way that a descending pass overflies a small islet called Lampione, 20 km west of Lampedusa. During the TOPEX/POSEIDON verification phase, until mid-December 1992, the satellite passed nine times over Lampedusa. The Lampedusa site was equipped with all the required instrumentation to provide independent and accurate knowledge of the altimeter system components: sea level, orbit, and geophsyical corrections. In addition, a reprocessing of the altimeter data was performed to better adjust the waveform parameters estimation especially when quasi-specular echoes due to flat sea conditions have disturbed the on-board estimation. Three different local orbits computed by expert groups (Centre d'Etudes et de Recherches Geodynamiques et Astronomiques, Delft University of Technology, and Jet Propulsion Laboratory), plus CNES and NASA global orbits were considered. The bias of the CNES altimeter (SSALT) was then calculated, based on six overflights, giving a mean value of 0.7 cm. A SSALT bias also was calculated at Harvest, the NASA calibration site, using six overflights. The deduced bias is 1.3 cm. Integrating both site results gives a mean SSALT bias of 1 cm with an uncertainty of 2.4 cm. The NASA altimeter (ALT) bias computed at Lampedusa is -18.5 cm with an uncertainty of 3.4 cm (the negative sign meaning that the altimeter is measurng short), but this result must be carefully considered because it was obtained with only two overflights. Considering the Lampedusa SSALT calibration results and those derived from statistical repeat track analysis giving an estimation of the relative bias between SSALT and ALT (e.g., Le Traon et al., this issue) a bias of -14.8 cm with an uncertainty of 2.6 cm was found for the ALT altimeter. ¿ American Geophysical Union 1994