An algorithm is developed which uses six channels of the Scanning Multichannel Microwave Radiometer to estimate fractions of multiyear and first-year sea ice types under variable atmospheric and sea surface conditions. Previously published sea ice algorithms have used a minimal number of channels and have therefore avoided most problems associated with noisy data as well as detailed atmospheric correction. Algorithms used for retrieval of atmospheric parameters concentrate on a statistical approach utilizing as many channels as possible. The present work attempts to synthesize these two classes of algorithms by using several radiometer data channels. This algorithm is specifically tuned to derive sea ice parameters while accepting error in the auxiliary parameters of surface temperature, ocean surface wind speed, atmospheric water vapor, and cloud liquid water content. The four additional environmental parameters are derived only for the purpose of correcting the sea ice parameters at the expense of accepting absolute errors in the derivation of the secondary parameters. These large errors propagate small errors in the resultant sea ice fractions. This weather-correcting algorithm performs at least as well, if not better, than the first-generation algorithms over consolidated ice and appears to perform very well when the sea ice concentration is less than 20%. To this end, the results are compared with the results of two algorithms that appear in the published literature. Unlike one of the algorithms that employs a weather filter, this new algorithm utilizes the radiative transfer equation to correct for these effects and thus eliminates false retrievals of sea ice without discarding data. The penalty that is paid is that more computer resources are required. ¿American Geophysical Union 1987