Based on the analysis of the quality level in a Pathfinder 4km daily nighttime Sea Surface Temperature product (PFSST) in the East China Seas (ECS) from 1985 to 2004, the proportion of high-quality data was lower than that in the global level. Additionally, the PFSST maps showed clearly the void and anomaly data impacted by atmospheric contamination. In order to solve the above problem, an optimal algorithm was established through introducing the structure function, setting up the daily first-guess sea surface temperature (SST) field and taking PFSST product of the highest quality as reference points. Comparisons were done between this optimally interpolated SST and the selected original PFSST and the simultaneous in situ measurements. It illustrated it was possible to exactly estimate the SST values in the ECS during the recent two decades. The mean bias error and the root mean square error between data sets optimally evaluated and in situ observed were lower than those between the previous global estimations and in situ measurnments.