The GSFC98 mean sea surface (MSS) was computed on a 2' oceanwide grid for latitudes below 80¿. The data used included 3-years of TOPEX data (Cycles 9 to 119), 1.5-years of ERS-1 35-day repeat cycle (Cycles 1 to 18), 2-years of Geosat ERM data (Cycles 1 to 42), 2 ERS-1 168-day repeat cycles, and 18 months of the Geosat Geodetic Mission data. All non-TOPEX satellite altimeter data were adjusted to the mean of TOPEX data in 2¿¿30¿ blocks. After the adjustment, the mean sea surface height was gridded into 2' nodes using least squares collocation and an iteration procedure to reduce the ocean variability. To validate the mean sea surfaces, three comparisons were made. The GSFC98 MSS, along with OSU95 [Yi, 1995> and CSR95 [Kim et al., 1995>, were compared with 6-years of TOPEX and 3-years of ERS-2 mean tracks which were not used in the MSS computations. Finally, the marine gravity anomalies were computed from the three MSS implied geoid undulations using the inverse Stokes integral. The marine gravity anomalies were compared with ship gravity data in selected areas. The ship gravity comparison is an independent assessment of the quality of the MSSs, especially at intermediate and short wavelengths. Finally, the inter-comparisons between the mean sea surfaces were made. The root mean square values of the differences were 6.8, 6.8, and 7.2 cm between GSFC98/OSU95, GSFC98/CSR95, and OSU95/CSR95. The differences agree well with the error estimation of GSFC98 MSS. ¿ 2000 American Geophysical Union