The new generation of GOES satellites have enabled accurate measurements of sea surface temperature (SST) from geostationary platforms. These satellites provide a significant increase in the number of cloud-free observations and complete observation of the diurnal cycle of SST. The changing solar geometry and heating of the satellite, however, can lead to significant variations in the bias in retrieved SST with time of day if the effects are not properly treated. We have integrated SST observations from GOES-8 and GOES-9 with more traditional SST observations from the AVHRR and buoys to produce accurate SST fields around the Americas every three hours. Overlapping coverage regions of GOES-8 and GOES-9 are used to cross-calibrate the two satellites and remove satellite-heating related variations in the brightness temperatures with time of day. Matchups between the GOES brightness temperatures and drifting and moored buoy measurements of SST are used to derive SST retrieval algorithms and verify their accuracy. Further comparisons with operational MCSST fields produced from AVHRR measurements are used to ensure the accuracy of the GOES-derived SST values in regions where in situ data coverage is not sufficient. The detailed procedure provides a reduction in the hourly variation in the SST bias from near 0.8 K in simple, single-satellite, GOES SST retrieval algorithms to approximately 0.3 K. The remaining variation is consistent with the natural variability in the bulk-skin temperature difference. The resulting product provides increased data coverage and accurate depiction of diurnal to seasonal variations in SST in the Americas and should prove very valuable to ongoing climate studies and weather forecasts. Variations in the SST related to the 1997/1998 ENSO cycle are highlighted in particular.