A new algorithm for estimating sea surface temperature (SST) from ERS-1 ATSR data using moisture dependent coefficients has been proposed. The clear-column water vapour (WV) information required for selecting the appropriate coefficients has also been derived from ERS-1 ATSR data. The method has been tested with the simulated ATSR data by comparing the estimated SST with the true SST used for simulating the data. The comparison of SST estimation errors by present and standard methods, under different noise conditions, has indicated improvements by the present method over the standard method. The small errors anticipated in SST estimation due to the errors in WV estimation causing erroneous selection of coefficients have also been taken care of by making use of overlapping ranges of WV for moisture dependent SST estimation coefficients. The method could be used for operational SST from ATSR data.