The TRMM Microwave Imager (TMI) was launched in November 1997 into an equatorial orbit that provides satellite coverage from 40N to 40S. At microwave frequencies, the frequent clouds of the tropics are not opaque and the radiative transfer based SST retrieval algorithm precisely accounts for their effect on brightness temperatures. Virtually complete coverage is assured in three days with only small areas of missing data due to persistent rain. When compared to the Tropical Atmosphere Ocean (TAO) buoy array SSTs, the microwave SSTs are shown to have a mean bias of -0.08°C and standard deviation of 0.57°C, an impressive result given that the buoy SSTs are not utilized during algorithm development. These attributes make TMI especially valuable for measuring SSTs during severe storms. Using 3 years of data, with over 50 severe storms, we analyze correlations between errors in the SST initialization field and forecast accuracy. As expected, when low vertical shear is present, many storms exhibit a strong correlation between SST and intensity. The strong control SST exhibits on intensity can cause errors in the initialization field to translate directly into forecast errors.