Presented at the 2017 International Geoscience and Remote Sensing SymposiumThe objective of this study is to investigate the effect of sea surface temperature (SST) on L-band microwave measurements and its implication on sea surface salinity (SSS) retrieval. Of particular interest is in the cold and fresh water where large SSS retrieval errors exist in comparison with Argo data. We found systematic SST dependence in Aquarius radar backscatter σo and radiometer excess emissivity Δe, with the emissivity of specular surface estimated using collocated HYCOM SSS and NCEP or SSMI/S wind. In the cold water under medium to high wind, σo and Δe show opposite trend on SST: σo reduces while Δe enhances by more than 10% relative to their corresponding values at the reference SST (15°C). The geographical distribution of matchups with SST<;5°C & wind speed between 5-20 m s-1 show that data collected under these conditions are coincident with the area where dSSS (=SSS-SArgo) is largely positive. This is consistent with the SST trend observed in σo, which would result in overestimation in roughness if the SST effect not considered. However, the enhanced Δe in cold water is puzzling because it would cause even higher value for SSS retrieval if modeled as roughness. SSS retrieved with SST correction reduces bias but results on error improvement is mixed. We hypothesis there is an unknown defect in the dielectric constant model for cold water and propose an empirical correction for the Aquarius SSS retrieval.