The Geosat (geodetic satellite) altimeter transmits a radar pulse toward Earth, then measures Earth-reflected energy in a number of time gates. The signal amplitudes in successive gates create a ''waveform,'' the shape of which is determined by the reflectance characteristics of the Earth's surface. This study shows that over the polar ice cap, the type and concentration of sea ice can be related to Geosat return waveform shapes. Because it is known that an altimeter waveform may be a mixture of returns from two or more ice types within the footprint, linear umixing theory is chosen as the framework for analysis of waveform data. The mixing model assumes that data samples are linear combinations of a limited of end members. Unmixing analysis therefore consists of identifying the number and nature of the end members, followed by an interpretation of the data set in terms of combinations of these end members. Estimates of ice type and concentration obtained by linear unmixing of waveforms are compared with ''ground truth'' derived from visual interpretation of passive microwave imagery and aerial photography, which were collected coincident with Geosat overpasses during the 1987 Marginal Ice Zone Experiment. Generally good agreement is observed between waveform-based ice information and the corresponding imagery ground truth. This analysis has revealed some deficiencies in a current parameterization of waveform shape called the Geosat ice index. A new ice index formulation is proposed.