Sea surface temperatures (SST) obtained from the SEASAT scanning multichannel microwave radiometer (SMMR) are compared with in situ SST observations in the eastern and central tropical Pacific Ocean for late July 1978. Comparing pairs of SMMR and ship SST (injection temperatures), we find that the SMMR SSTs are biased 0.9¿C lower than ship SSTs, with a standard deviation (relative to the bias) of 1.7¿C. Since ship SSTs tend to concentrate along shipping lanes, while SMMR SST's are distributed nearly uniformly, the statistics based on data-to-data comparisons are spatially biased: in addition, there is ambiguity in comparing the spatially average (149 km ¿ 149 km) SMMR SST against the point-observed ship SST. These ambiguities in data comparison can be reduced by comparing the SST fields derived from these data. The result shows that the SMMR-derived SST field (TSMMR) is, on the average, lower than the ship-derived SST field Tship by 0.8¿C with a standard deviation of 0.8¿C. Adjusted for 0.2¿C to 0.7¿C warm bias of ship SST <Saur, 1963: Tabata, 1978>, this set of SMMR SST has a bias -0.6 to -0.1¿C. The difference between Tship and TSMMR shows a strong regional dependence, as found in comparing SST fields derived either from different data sets or from the same data set but with different methods <Barnett et al., 1981>. In our case, the standard deviation of the difference between these two fields is found to be inversely proportional to the spatial density of ship data. The implications are that the spatial variation of the SST field may be beter derived from SMMR SST than from ship SST, which is handicapped by its nonuniform data distribution, and after determining the biases of both SMMR and ship SST, these two data sets may complement each other for constructing SST fields.