Cross-spectral analysis of repeat satellite altimeter profiles was performed to compare the along-track resolution capabilities of Geosat, ERS 1 and TOPEX data. Geophysical Data Records were edited, differentiated, low-pass-filtered, and resampled at 5 Hz. All available data were then loaded into three-dimensional files where repeat cycles were aligned along-track (62 cycles of Geosat/Exact Repeat Mission; 16 cycles of ERS 1, 35-day orbit; 73 cycles of TOPEX). The coherence versus wave number between pairs of repeat profiles was used to estimate along-track resolution for individual cycles, eight-cycle-average profiles, and 31-cycle-average profiles (Geosat and TOPEX only). Coherence, which depends on signal to noise ratio, reflects factors such as seafloor gravity amplitude, regional seafloor depth, instrument noise, oceanographic noise, and the number of cycles available for stacking (averaging). Detailed resolution analyses are presented for two areas: the equatorial Atlantic, a region with high tectonic signal and low oceanographic noise; and the South Pacific, a region with low tectonic signal and high oceanographic variability. For all three altimeters, along-track resolution is better in the equatorial Atlantic than in the South Pacific. Global maps of along-track resolution show considerable geographic variation. On average globally, the along-track resolution (0.5 coherence) of eight-cycle stacks are approximately the same, 28, 29, and 30 km for TOPEX, Geosat, and ERS 1, respectively. TOPEX 31-cycle stacks (22 km) resolve slightly shorter wavelengths than Geosat 31-cycle stacks (24 km). The stacked data, which are publicly available, will be used in future global gravity grids, and for detailed studies of mid-ocean ridge axes, fracture zones, sea mounts, and seafloor roughness. ¿ American Geophysical Union 1995