It is demonstrated that an along-track mean sea surface (MSS) model estimated with TOPEX altimeter data, including the large 1997-1998 El Niño event, is slightly less accurate than a MSS model calculated from less data where El Niño signals are small. The manner in which true sea level variability corrupts the estimation of MSS gradients is discussed. A model is proposed to reduce the error, based on scaling climate indices such as the Southern Oscillation Index, while accounting for phase shifts using a Hilbert transform. After modeling and removing the seasonal and interannual sea level variations, parameters to a plane MSS model are estimated using TOPEX altimeter data from January 1993 to June 2000. Results indicate an overall improvement over the earlier model based on four years of data, and no apparent degradation due to aliasing of sea level variability.