An algorithm is currently under development that will provide a classification mask for ASTER imagery obtained poleward of 60 N and 60 S. The classification mask will be a product available through EOSDIS and is called the ASTER polar cloud mask. Ten classes are currently in the mask and include six clear classes (water, slush/wet ice, ice/snow, land, shadow on land, and shadow on ice/snow) and four cloud classes (thin cloud over ice/snow, water, or land, and thick cloud). The algorithms is designed as a four stage process. In the first stage the data are median filtered, sampled to 30 m spatial resolution, normalized, and navigated to coastlines and ancillary Earth surface databases. In the second stage, through adaptive thresholding, simple decision surfaces, and ancillary data, the class ambiguity of each pixel is reduced from ten to two to four classes. In the third stage, additional features are utilized in a paired- histogram classification methodology to make the final pixel classification. And finally, in the fourth stage, a simple spatial consistency check is performed over the entire classification mask to detect isolated pixel classifications. Over 3700 samples have been extracted and labeled to date representing over one million pixels from 82 Landsat TM circumpolar scenes. Tests of the algorithm on the labeled samples indicate that the clear/cloud classification accuracy is greater than 90 percent and subjective evaluation of the classification masks supports that result.