A method of analyzing salinity, as well as temperature, is adopted in the equatorial Pacific. It is a three-dimensional variational method with vertical coupled temperature-salinity empirical orthogonal function modes. The salinity field is estimated from temperature observation alone such that the surface dynamic height calculated from the estimation has realistic variability compared with TOPEX/Poseidon (T/P) altimetry data. Using T/P altimetry data with temperature observation further improves the analysis of the salinity field. The comparison with the in situ salinity data showed that the salinity fields were analyzed successfully with temperature and T/P observation data alone especially in the western equatorial Pacific, although there are some difficulties in analyzing salinity in the eastern equatorial Pacific and the surface mixed layer. These results convince us that the analysis salinity field has its own reliability even where no salinity observation exists. We also confirm that interannual variabilities of the near-surface salinity field and the barrier layer in the period of 1993--2000 in the analyses are consistent with former studies. The fresh water is confined to the west, sea surface salinity in the central equatorial Pacific is relatively high, and a thick barrier layer develops in the western equatorial Pacific during La Ni¿a periods. The fresh water spreads to the central equatorial Pacific, and the thick barrier layer moves to the east during El Ni¿o periods. The correlation between near-surface temperature and the barrier layer thickness is also confirmed. The analysis fields presented here are quite adequate to study the salinity interannual variability.