Recent advances in the area of satellite altimetry have provided a data type which is applicable to the detection and monitoring of transient sea surface phenomena. With the extensive altimeter data set collected by the Geos 3 spacecraft there is an associated need for accurate and efficient methods for editing and filtering the altimeter data. In this investigation a sequential filter is developed assuming that the sea surface topography can be approximated by an adaptive Gauss-Markov process. The effect of geoid model errors on the filtered results is illustrated by using both 1¿¿1¿ and 5'¿5' geoid models. Finally, an application of the method to the Gulf Stream current boundary identification problem using Geos 3 altimeter data is described.